eMule AI modernizes the eMule experience with NAT Traversal, QUIC/uTP and IPv6 support, stronger LowID and firewalled peer connectivity, smoother interface performance and a broad range of usability, protection and reliability improvements.
This page brings together the key information around the project: an overview of the client, technical guidance for advanced connectivity features, release notes for recent work, and support sections that help users get started and troubleshoot effectively.
Connect more peers through NAT Traversal, including "eServer assisted LowID to LowID" and "Kad firewalled to firewalled" transfers, with QUIC or uTP used for direct client to client traffic.
Use the official emuleai/emuleai Docker image on Windows, Linux, or macOS with the release launcher package. Xpra presents eMule AI as a separate desktop window while noVNC provides browser based VNaccess.
Background loading, virtual lists, shared file caching and lighter bulk operations keep startup, searches, transfers and very large libraries smooth and responsive.
A complete dark theme, refined dialogs, clearer progress overlays and responsive layouts make everyday use cleaner and easier to follow.
Centralized anti-leecher and abuse controls let you choose fair actions for suspicious behavior with timed rules, friend exemptions and safeguards against false positives.
Download Validator, Download Inspector, spam detection and blacklist tools help avoid duplicates, misleading files and unwanted results before they become a problem.
Flexible slot targets, idle slot recycling, adaptive buffering and queue controls help make better use of both everyday and high-bandwidth connections.
Background scanning, folder watching, startup caches and automatic subdirectory sharing keep large collections current with less manual work.
VPN Guard, interface binding, IPv6 support, firewall rule management and optional port randomization provide clearer control over how eMule AI connects.
Configurable eD2K and Kad limits, fast filters, restored search tabs and clearer file states make large result sets easier to explore and manage.
A guided migration wizard, automatic backups, safer state saving and restored source data help protect your setup and resume interrupted work quickly.
Useful toolbar actions, richer sortable columns, client notes, flexible notifications and 116 built-in languages make common tasks faster and more accessible.
This section follows the Options window from top to bottom. Each page and setting appears in the same order as the application and uses the same explanation shown in its tooltip.
Enter the name shown to other eMule users. It does not need to be unique, but a recognizable name is easier for friends to identify.
Choose the language used by eMule menus, dialogs and messages. The interface updates after you apply the change.
These settings control how eMule reacts to links, closing, status reporting and the MiniMule window.
Bring the eMule window to the front when an eD2K link is opened. When disabled, eMule accepts the link without taking focus from the application you are using.
Ask for confirmation before eMule closes. This helps prevent an accidental exit that would stop all transfers and network connections.
Write current connection and transfer information to onlinesig.dat for use by external tools or web pages. The file may reveal your server address and transfer activity.
Enable the compact MiniMule status window. Minimize eMule to the notification area then click its tray icon to view connection status, speeds, completed files and free space.
Keep Windows from entering automatic standby while eMule is connected so uploads and downloads are not interrupted. Manual sleep commands and display power saving still work.
Open the Web Services editor to add, edit, remove or reorder the web searches and links available from file context menus.
Register eMule as the default application for eD2K links so links opened from a browser or another program are sent to eMule.
Choose how eMule looks and behaves when it starts with Windows or when you launch it yourself.
Show the eMule splash image while the application is starting.
Start eMule minimized instead of opening the main window. Its icon remains available in the notification area or taskbar according to your Display settings.
Launch eMule automatically after you sign in to Windows. Disable this when you prefer to start eMule manually.
Select a larger Options window.
Set how long the pointer must remain over a supported control before its tooltip appears. The value is measured in seconds.
Change the shape of progress bars from flat to rounded. This affects their appearance only and does not change transfer behavior.
Send eMule to the Windows notification area when it is minimized instead of leaving a button on the taskbar.
Require a double click to expand or collapse a download and view its sources. When this is disabled, the normal expansion control responds immediately.
Start a preview when you double click the file name of a previewable download. The file must contain enough completed data for previewing.
Display the completed percentage inside each download progress bar so progress can be read without opening file details.
Add the current upload and download speeds to the eMule window title so transfer activity remains visible on every page.
Show the number of active downloads and the total number of files on each category tab.
Remove completed entries from the Transfers list automatically. The downloaded files remain in the Incoming folder and continue to follow your sharing settings.
Show the additional toolbar in Transfers for filtering downloads and accessing common file actions.
Save open search tabs and their current results when eMule closes then restore them at the next start.
Use supported Windows taskbar features to show combined download progress and preview controls on the eMule taskbar button.
Display helpful explanations when you rest the pointer over controls throughout Options. Turn this off if you prefer to use the pages without hover help.
Reduce memory and processor use by stopping lists that you do not need. A disabled list is not displayed until you enable it again.
Stop maintaining the Known Clients list to reduce memory use. The list is unavailable until this option is turned off again.
Stop updating the visible upload queue list to reduce processor and memory use. Uploading continues but queued clients are not shown.
Choose the font, style and text color used in the Server, Messages and IRC windows.
Store previous search terms and address entries then suggest them while you type. Disable this if you do not want eMule to keep this history.
Delete all saved autocomplete entries. This action cannot be undone.
Enter the real maximum speeds of your internet connection. Accurate upload capacity is especially important because eMule uses it to manage bandwidth without blocking normal connection traffic.
Set optional speed limits for eMule. Leaving a limit disabled allows eMule to use the available bandwidth while other connection controls remain active.
Enter the highest download speed your internet connection can provide. The value uses Mbit/s or KB/s, which matches the unit currently shown beside this field. This describes your connection capacity and does not limit eMule by itself.
Limit the download bandwidth eMule may use. Enable this when other applications need part of the downstream capacity or leave it disabled for no download limit.
Enter the highest upload speed your internet connection can provide. The value uses Mbit/s or KB/s, which matches the unit currently shown beside this field. Enter the real line rate, or zero to let automatic upload management estimate it.
Limit the upload bandwidth eMule may use. Keep some upload capacity free for acknowledgements and protocol traffic or download speed and web browsing may suffer.
Configure the listening ports used for incoming peer connections. Router forwarding and firewall rules must match these values to obtain High ID and an open Kad status.
Set the TCP listening port used for direct eD2K client connections and file transfers. TCP establishes a reliable connection before data is exchanged. Forward this port through the router and allow it through the firewall.
Set the UDP listening port used by Kad and by low overhead messages between clients and servers. UDP sends individual datagrams without a persistent connection. Forward this port for an open Kad status.
Disable eMule UDP communication when this port cannot be used. Kad will be unavailable and more client communication must use TCP, which increases overhead.
Open the eMule port test in your browser to check whether the configured TCP and UDP ports accept incoming connections. A failed test usually means the firewall or router settings need attention.
Let eMule ask a compatible router to forward the configured ports automatically at startup. Use manual port forwarding if UPnP is unavailable or unreliable.
Create or update Windows Firewall rules for the current TCP and UDP listening ports so incoming connections are not blocked on this computer.
Choose new TCP and UDP listening ports from the specified range each time eMule starts. Your firewall and router must allow the full range.
Define the port range used when random listening ports are enabled. The router and firewall must allow the complete range because a different value can be chosen at each start.
Enter the first port that eMule may choose for randomized TCP and UDP listening ports.
Enter the last port that eMule may choose for randomized TCP and UDP listening ports. It must not be lower than the start value.
Set the maximum number of sources eMule keeps for each download. A very high value increases memory use and network requests without always improving speed.
Set the maximum number of simultaneous network connections eMule may open. Values beyond the operating system and router capacity can cause instability or connection failures.
Connect automatically to the enabled eD2K and Kad networks when eMule starts.
Choose which peer to peer networks eMule may connect to. At least one network must be enabled to find sources and exchange files.
Try to restore a server or Kad connection automatically after it is lost.
Allow eMule to connect to Kad, the decentralized network that finds files and sources without relying on a central server. Kad requires UDP for full operation.
Include protocol overhead in the displayed upload and download rates. Overhead is connection management data rather than file content.
Allow eMule to connect to the eD2K server network. Servers help publish shared files and locate sources but file data is exchanged directly between clients.
Display all transfer speeds in KB/s instead of automatically switching between KB/s and MB/s.
Open the connection wizard to fill common capacity and limit values from a known internet connection type. Review the result before applying it.
Choose the network adapter or local IP address that eMule AI should use for P2P traffic.
Select the network adapter for P2P sockets. Choose Any interface to let Windows route normally.
Optional local IPv4 address to bind. Leave empty to use the current address of the selected adapter.
Enable IP Guard. While enabled, eMule AI blocks network activity until the selected adapter/IP and optional public IP ranges are verified. If verification fails, eMule AI stays open so you can fix the settings.
Optional. Enter allowed public IPv4 or IPv6 addresses and CIDR ranges for IP Guard, separated by commas. Example: 1.2.3.4/24, 2001:4860::/32
Shows the adapter and local IP address currently used for P2P binding.
VPN Guard prevents eMule from connecting when the public IP appears in an unsafe location. It helps avoid accidental exposure when a VPN disconnects or uses an unexpected exit location.
Enable VPN Guard and block eMule network access while the public IP is detected in the selected country.
Choose the country that VPN Guard treats as unsafe. eMule blocks its network connections while the public IP is detected there.
Also block eMule when the country of the public IP cannot be determined. This is safer but may pause connections during temporary lookup failures.
Configure a proxy only when the network administrator or provider requires one. A home router or modem is not a proxy and incorrect values can prevent every eMule connection.
Route supported eMule TCP connections through the configured proxy server. Most proxy types do not support UDP, so Kad and UDP features may need to be disabled.
Choose the protocol offered by the proxy server. The selected type must match the server configuration or connections will fail.
Enter the host name or IP address of the proxy server supplied by the network administrator or proxy provider.
Enter the TCP port used to connect to the proxy server. This is the proxy service port, not the eMule listening port.
Use these settings only when the proxy server requires a user name and password.
Send the configured credentials when connecting to the proxy. Leave this disabled for proxies that allow access without authentication.
Enter the user name required by the proxy server.
Enter the password required by the proxy server. The value is hidden on screen but is stored with the eMule preferences.
Control how eMule obtains and maintains its eD2K server list. Use trusted server list sources because invalid entries can reduce connection quality.
Set how many failed status checks are allowed before an unresponsive server is removed from the list.
Update the server list at startup from the first working address stored in addresses.dat.
Open addresses.dat to edit the server list update addresses used at startup. Save the file before the next update.
Add server entries supplied by the eD2K server after a successful connection. Disable this when you prefer to maintain a trusted list manually.
Add server entries exchanged with other eMule clients. This can expand the list but may also introduce outdated or unwanted servers.
These settings control Low ID handling, connection safety and server priority.
Treat an unexpected Low ID as a possible server problem when this computer previously obtained High ID. Repeated Low ID results are accepted after several attempts.
Contact one server at a time during connection attempts. This is slower but can reduce incorrect Low ID results on some networks.
Connect only to servers stored in staticservers.dat. eMule cannot connect if the static list contains no working server.
Prefer servers according to the priority assigned in the Server list.
Assign high priority automatically to servers you add manually.
Choose the folder where completed downloads are moved. Files in this folder are normally shared according to the current sharing settings.
Choose one or more folders for incomplete part files. Use disks with enough free space and avoid moving or editing these files while eMule is running.
Add another temporary folder. eMule can distribute new part files across the configured locations according to available space.
Select the folders that eMule publishes to other users. Hold Ctrl while clicking where supported to include or exclude subdirectories and avoid sharing private folders.
Add a shared folder by its UNC network path, such as a folder on a NAS or another Windows computer. The path must remain reachable while eMule is running.
Choose the initial state and priority assigned when new downloads or shared files are added.
Add new downloads in paused state so they do not start transferring until you resume them manually.
Choose upload priority automatically for newly shared files according to demand and availability.
Choose download priority automatically for new downloads according to source availability and queue conditions.
Clean common release separators and unwanted text from new download names before they are shown and saved.
Open the file name cleanup rules used for new downloads. Review the rules carefully because they can change the saved file name.
These settings control chunk selection, completion order, history and other download behavior.
Try to send complete file chunks to upload clients instead of frequently switching between chunks. This may improve chunk distribution but can keep one client active longer.
Prefer the beginning and end chunks needed by many media formats so a partial download may become previewable sooner.
Watch the Windows clipboard for eD2K links and offer to add them when copied from another application.
Use the advanced calculation for estimated download completion time. It reacts to recent transfer behavior but the result is still an estimate.
Resume another paused download automatically when a file completes.
Prefer the next paused file from the same category as the completed download.
Resume a paused file only when one exists in the same category. Other categories remain paused.
Keep completed file hashes in known.met so Download Validator and duplicate checks can recognize files that were downloaded before.
Keep hashes of cancelled downloads so eMule can recognize them later and warn or reject them according to Download Validator settings.
Configure the external program used to preview video downloads and the command line passed to it.
Enter or browse to the executable used for video preview. Leave this empty to use the Windows default application when supported.
Enter optional command line arguments for the video player. The file name is added according to the preview command handling.
Create a temporary backup before previewing a partial video. This is safer for players that modify the file but requires additional disk space and time.
Choose how desktop notifications are displayed. The available modes depend on the notification support provided by the current Windows version and eMule theme.
Choose how eMule alerts you when a notification is shown.
Show notifications without playing or speaking an alert.
Show a test notification using the current display and sound settings without saving them first.
Play the selected wave file when a notification is shown.
Enter or browse to the wave file used for notification sounds.
Read notification text aloud through the Windows speech service instead of playing a wave file.
Choose which events create a desktop notification. Enabling many frequent events can produce a large number of popups.
Show a notification when a new entry is written to the eMule log.
Show a notification when another user starts a chat session.
Show a notification whenever a chat message is received, including messages in an existing session.
Show a notification when a new download is added to the transfer list.
Show a notification when a download completes successfully.
Show notifications for important problems such as low disk space or a lost server connection.
Send selected notifications by email. A working SMTP configuration and valid sender and receiver addresses are required.
Enable email delivery for eMule notifications after the SMTP server and addresses have been configured.
Open the SMTP server settings used to send notification email, including the server address, port and authentication details.
Enter the email address that should receive eMule notifications.
Enter the sender address placed in notification emails. Some SMTP servers require this address to match the authenticated account.
Configure how the Statistics graphs are updated, scaled and colored. Faster updates use more processor time.
Set how often the live Statistics graphs are refreshed. A shorter interval looks smoother but requires more processor time.
Set the time period represented by the average graph. A longer period smooths short speed changes and shows broader trends.
Choose a graph element then select the color used to draw it.
Open the color picker for the graph element selected in the list.
Fill the area below graph lines instead of drawing lines only. This changes appearance and may make overlapping values harder to compare.
Set the vertical scale used by the connection statistics graph. Increase it when high connection counts are clipped at the top.
Choose the ratio used to display active connections in the connection statistics graph.
Configure how often values in the Statistics tree are recalculated and redrawn.
Set how often the Statistics tree is refreshed. A longer interval reduces processor use on large or busy sessions.
Enter the IRC server that eMule should connect to. Use a host name supplied by the IRC network and include a port only when the server requires a nonstandard one.
Enter the nickname shown to other IRC users. It may need to be different from your eMule user name and must follow the rules of the IRC network.
Filter the channel list by name and minimum user count so large public channel lists are easier to browse.
Apply the channel name and minimum user filters when eMule receives the IRC channel list.
Enter text that a channel name must contain to remain visible in the filtered channel list.
Hide channels with fewer users than this number. Set a low value when you also want to see small channels.
Configure IRC commands that eMule sends automatically after connecting to the server.
Send the perform string automatically after a successful IRC connection.
Enter IRC commands to run automatically after connecting. Separate multiple commands with the | character. Open the Help tab for command syntax and examples.
These options control IRC messages, links, sounds, text encoding and other chat behavior.
Join or make the official eMule help channel easy to access after connecting to IRC.
Request and display the IRC channel list automatically after connecting. Large networks may take time to return the complete list.
Add the local time to IRC messages so the order and arrival time remain clear in long conversations.
Choose which routine IRC status messages are hidden from the chat window. Hiding them reduces clutter but can conceal useful connection details.
Hide miscellaneous IRC server information messages.
Hide messages that announce when users join a channel.
Hide messages that announce when users leave a channel normally.
Hide messages that announce when users disconnect from the IRC network.
Hide low level IRC ping and pong keepalive messages. These messages are normally useful only for troubleshooting.
Choose which eMule specific IRC protocol messages are ignored. These messages support actions such as friend requests and file links.
Ignore eMule friend requests received through IRC.
Ignore eD2K file links sent through eMule specific IRC messages.
Allow IRC users to add you as an eMule friend through the supported eMule IRC protocol.
Allow supported eD2K links received in IRC to be opened or added through eMule. Review links from unknown users before downloading.
Accept actionable eD2K links only from users who are already in your Friends list.
Replace supported text emoticons with graphical smileys in IRC messages.
Play sounds for selected IRC events such as messages, joins or mentions according to the current notification resources.
Use Unicode text encoding for IRC messages. Enable this on networks that support UTF-8 so names and messages from different languages display correctly.
Control which private messages are accepted and how chat content is displayed.
Enter words or phrases that cause an incoming private message to be ignored. Separate multiple entries with the vertical bar character.
Accept private messages only from clients in your Friends list. Other users cannot start a normal chat with you.
Use additional checks to identify automated or abusive private messages. The filter may challenge unknown senders before accepting their message.
Require unknown message senders to complete a captcha challenge before their messages are accepted. Friends are not normally challenged.
Replace supported text emoticons with graphical smileys in chat messages.
Control filtering and indicators for file comments and ratings received from other users.
Enter words or phrases that cause a received file comment to be hidden. Separate multiple entries with the vertical bar character.
Show an icon on downloads that have received a file comment or rating so you can review the information before completion.
Use ipfilter.dat to block communication with listed IP ranges. A stricter filter can improve privacy but outdated or poor quality lists may block valid sources and servers.
Apply the IP filter to eD2K servers as well as clients. Filtered servers are not used for connections or updates.
Keep private LAN address ranges outside the IP filter. Enable this when eMule must communicate with trusted devices on the local network.
Block IP filter entries whose access level is equal to or below this value. The common default is 127, but the correct value depends on the filter list format.
Reload the current IP filter file from disk and apply changes that were made outside eMule.
Open the current IP filter file in the default text editor. Save your changes then reload the file to apply them.
Enter the address of an ipfilter.dat or compatible PeerGuardian list. Use a trusted source because the downloaded rules can block all matching network traffic.
Open the saved history of IP filter update addresses and choose one for the current update.
Download the IP filter from the address entered above and apply the updated filter entries.
Download the configured IP filter automatically at the selected interval.
Set the number of days between automatic IP filter updates. Very frequent updates are unnecessary unless the provider changes the list often.
Choose who may browse the list of files and directories shared by this eMule client. This does not stop normal search results or file transfers.
Allow any connected eMule user to request your shared file list.
Allow only users in your Friends list to request your shared file list.
Do not provide the complete shared file list to other users. Individual files can still appear in normal eD2K and Kad searches.
Protocol obfuscation makes eMule traffic less obvious to simple traffic shaping systems. It is not encryption and does not hide your IP address or activity from peers.
Allow outgoing and incoming obfuscated connections when the other client supports them while retaining compatibility with normal connections.
Accept only obfuscated client connections. This reduces the number of reachable sources because older or incompatible clients cannot connect.
Disable all support for protocol obfuscation and use normal eD2K connections only.
These settings control client identity checks, application privileges, search spam filtering and warnings for risky files.
Use cryptographic secure identification to protect the credit relationship between clients and reduce credit theft by impersonation.
Run eMule with reduced Windows privileges to limit the effect of a malicious file or software flaw. This option may be unavailable on unsupported systems.
Detect and mark suspicious search results using file names, metadata, client behavior and configured blacklist rules.
Warn before opening file types or locations that may be unsafe. The warning does not scan the file for malware, so use trusted security software as well.
Enable the scheduler so active entries can change eMule settings at their configured times. Disabled schedules remain saved but do not run.
Delete the selected schedule entry. The change is saved when you apply the Options page.
Create a new schedule entry then configure its name, time and actions in the Details section.
Select a saved schedule to view or edit its details. Each entry can run one or more actions on chosen days and times.
Configure the selected schedule name, recurrence, time range and actions.
Enable or disable the selected schedule entry without deleting its configuration.
Enter a descriptive name that helps identify the purpose of this schedule in the list.
Choose the days or recurrence pattern on which the selected schedule is allowed to run.
Set the time when the selected schedule begins applying its actions.
Set the time when the selected schedule stops or reverses actions that use an end time.
Run the schedule without a separate end time. This is useful for one time actions or settings that should remain active until another schedule changes them.
Choose the actions performed by this schedule. Actions can change upload and download limits, source and connection limits, network availability, category state or the eMule application state. TCP and UDP refer to network connection settings, not transfer speeds.
Apply the edited details to the selected schedule entry. Use the main Options Apply button afterward to save all pending changes.
Configure the Web Interface included with eMule to monitor and control the application from a browser. Exposing it outside the local network requires strong passwords, HTTPS and careful firewall rules.
Start the eMule Web Interface on the configured port. Anyone who can reach the port can attempt to sign in.
Compress web pages sent to the browser to reduce bandwidth. This uses a small amount of additional processor time.
Ask a compatible router to include the Web Interface port in the UPnP forwarding setup. This may expose the interface to the internet, so use HTTPS and strong passwords.
Set the TCP port used by the Web Interface. This port is separate from the eD2K listening ports and must be allowed by the firewall for remote access.
Choose the folder or template file set used to render Web Interface pages. Use templates designed for the installed eMule version.
Open a folder browser to choose the template files used by the Web Interface.
Reload Web Interface templates from disk without restarting eMule. Existing browser sessions may need to refresh the page.
Set how many minutes an inactive Web Interface login remains valid before the user must sign in again.
Protect Web Interface traffic with HTTPS. A valid certificate and private key are required to avoid sending passwords and commands as plain text.
Create a new local certificate and private key for HTTPS. Browsers may warn about a certificate that is not signed by a trusted authority unless you trust it manually.
Enter or browse to the certificate file presented by the HTTPS Web Interface.
Enter or browse to the private key that belongs to the HTTPS certificate. Keep this file private and restrict access to it.
Configure the administrator account, which has access to all Web Interface controls.
Set the password for the Web Interface administrator account. Use a long unique password, especially when remote access is possible.
Allow the administrator to close eMule or request operating system reboot and shutdown from the Web Interface. Enable this only on a trusted and protected interface.
Configure a limited Web Interface account for monitoring and lower risk actions without full administrator permissions.
Enable the limited Web Interface user account.
Set the password for the limited Web Interface account. Use a password different from the administrator password.
These advanced settings can reduce performance or connectivity when used incorrectly. Change them only when you understand the effect and keep a record of the original values.
Configure advanced TCP connection limits and server connection timing. Values that are too high can overload the router or operating system.
Set how many new TCP connection attempts eMule may start within five seconds. A lower value is safer for older routers and restrictive networks.
Set the maximum number of TCP connections that may still be waiting for completion. Excessive values can trigger operating system or router limits.
Accept incoming TCP connections conditionally when normal listening behavior cannot be used. Leave this at the default unless a specific network setup requires it.
Set how long eMule keeps an idle server TCP connection alive before sending maintenance traffic or reconnecting.
Reclaim the Windows eD2K link association automatically when another application changes it.
Use the eMule credit system, which rewards clients that have uploaded data to you with shorter queue waits. Disabling it removes this local credit advantage.
Reject sources that use private LAN addresses when they arrive from the public networks. Disable this only when you intentionally exchange sources inside a trusted local network.
Show additional expert commands and controls in eMule menus and dialogs. These actions are intended for experienced users.
Disable repeated A4AF checks to reduce processor use. A4AF manages sources that are available for another file, so disabling checks can make source reassignment less responsive.
Prevent eMule from automatically creating archive previews. Disable previews when archive inspection is slow or unnecessary.
Enter the host name eMule should advertise when a network feature needs one. Leave it empty unless the connection has a stable and correctly configured host name.
Store incomplete downloads as sparse files on supported NTFS volumes so unreceived areas do not consume physical disk space.
Reserve the complete final file size when a download starts. This detects low disk space early but may allocate large amounts of storage immediately.
Pause downloads when free disk space falls below the configured minimum. This protects the file system from becoming completely full.
Set the free disk space that eMule should preserve on a volume before pausing affected downloads.
Set how often eMule checks free disk space for active download locations. Frequent checks react sooner but add file system work.
Choose when eMule forces buffered file data to stable storage. More frequent commits reduce possible data loss after a crash but increase disk activity.
Do not force part file buffers to disk beyond normal operating system behavior. This minimizes extra disk activity but leaves more recent data dependent on system caching.
Commit buffered part file data when eMule closes normally. Unexpected power loss can still leave recent writes only in the operating system cache.
Commit part file data frequently for stronger persistence. This can cause considerably more disk activity, especially with many downloads.
Choose whether eMule reads media metadata such as title, artist, codec and duration from completed or shared files. Stored metadata can be reused in lists and searches.
Do not scan files for media metadata. Lists use only information already available from file names and saved tags.
Read supported audio and media tags with the metadata library included in eMule and save useful values for later display.
Follow Windows shortcut files when scanning shared directories. Disable this when shortcuts should be shared as files rather than resolved to their targets.
Write normal eMule log messages to a file in addition to showing them in the interface. Log files grow over time and may contain IP addresses or file names.
Configure detailed diagnostic logging. Use this mainly for troubleshooting because verbose logs can grow quickly and may expose private connection details.
Enable detailed debug logging for the selected event categories.
Choose how much diagnostic detail is recorded. Higher detail produces larger logs and more processing work.
Write verbose diagnostic messages to a debug log file so they remain available after eMule closes.
Record detailed source exchange requests and responses for troubleshooting source discovery.
Record when a client is banned and the reason that triggered the ban.
Record received file ratings and descriptions for troubleshooting metadata exchange.
Record secure identification handshakes and failures used by the client credit system.
Record connections and sources rejected by the IP filter. This can create many entries with large filter lists.
Record file save, rename and move operations performed when downloads complete or metadata changes.
Record source reassignment decisions for sources that are available for another file.
Record detailed upload and download state changes for troubleshooting transfers.
Record how search results are scored or marked by the spam detection system.
Record attempts to retry failed TCP connections and their results.
Record additional source exchange events beyond the standard diagnostic entries.
Record NAT traversal, callback and uTP related events used to connect clients behind routers or firewalls.
Record operations that may delay the user interface so freezes and slow list updates can be investigated.
Configure Upload Speed Sense, which adjusts upload bandwidth by measuring connection delay. Good values keep the connection responsive while using available upload capacity.
Enable Upload Speed Sense to adjust the upload limit automatically according to measured network delay.
Set the lowest upload speed that Upload Speed Sense may choose while responding to high network delay.
Set the allowed increase in ping time as a percentage of the measured baseline before upload is reduced.
Set the allowed increase in ping time in milliseconds before Upload Speed Sense reduces upload.
Choose whether Upload Speed Sense measures allowed delay as a percentage of baseline ping or as a fixed number of milliseconds.
Use a percentage of the baseline ping as the allowed delay. This adapts to connections with different normal latency.
Use a fixed millisecond value as the allowed delay. This gives a consistent threshold when normal latency is stable.
Control how cautiously Upload Speed Sense raises the upload limit when latency is healthy. Larger divisors generally make increases slower.
Control how strongly Upload Speed Sense lowers the upload limit when latency becomes too high.
Set how many ping measurements are considered when Upload Speed Sense evaluates connection delay. More samples are steadier but react more slowly.
Configure advanced UPnP behavior for automatic router port forwarding. These options affect which router services eMule tries and how mappings are removed.
Remove eMule port mappings from the router when the application closes normally.
Skip the router WANIPConnection service during UPnP setup. Use this only when that service is broken and another supported service works.
Skip the router WANPPPConnection service during UPnP setup. Use this only when that service causes errors or incorrect mappings.
Allow eMule to import usable data from compatible incomplete files into an active download. Imported data is verified before it is accepted.
Set how long completed writes may remain in memory before eMule flushes them to disk. Longer values can reduce disk activity but increase memory use and the amount of unwritten data lost after a crash.
Set the memory reserved for buffering file writes. A larger buffer can reduce disk activity but consumes more memory and gives little benefit once the storage device is saturated.
Set the maximum number of clients kept in the upload waiting queue. A very large queue consumes more memory and does not increase the available upload bandwidth.
The settings below and the preferences.ini button provide access to advanced behavior that is not normally needed for everyday use.
Open preferences.ini in the default text editor for advanced settings that are not available in Options. Close eMule before editing the file so your changes are not overwritten.
Here you can find tweaks related to user interface.
You can turn on/off dark mode here.
Sychronizes eMule AI's dark mode with your Windows setting.
Enables dark mode.
Enables light mode.
Display speed graphics in toolbar
Shows the vertical download toolbar in the transfers window. Disable this option to hide it and reclaim space immediately. (default: checked)
"Find As You Type" is an official feature that starts filtering files immediately as you type on the lists. When you disable "Find As You Type", filtering will be started after an enter or return key press, when you complete typing your text filter.
This is the time interval in milliseconds that will be waited before updating items in lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
You can change the maximum number of sort order history for the specified list here. eMule AI can sort orders of lists by multiple number of colums by remembering user clicks on columns. So if the data of two rows in the last clicked columns (marked with arrow) are equal then eMule AI uses next columns in the sort order history. Default value is 4. 0 means unlimited number of sort order history. Please note that using 0 or big values can effect user interface performance negatively if you have too much items in your download/upload/shared file/file history lists.
Provides information about the location of a server / client using DB-IP.com's IP to City Lite database.
Disable IP geolocation
This option shows only country code, doesn't show country or city names.
This option shows country name, doesn't show city name.
This option shows country and city names
Show country flags
Automatically download and reload the IP geolocation database.
URL template used to download the DB-IP City Lite MMDB archive. %Y and %m are replaced with the current UTC year and month.
Number of days between automatic IP geolocation update checks.
Downloads the configured IP geolocation database now.
Here you can find tweaks related to connection.
When activated, eMule AI constantly checks the internet if it is still connected, disconnects the Server/Kad connection when going offline and reconnects when going online. This prevents losing sources, Kad nodes or server entries when the connecion is offline. This implementation uses a native Windows API to check if the given URL address is reachable or not, instead of using ICMP which can be troublesome with some modem/router configurations. Enabling this option is recommended. Default: False
Toggle the connection checker. When enabled, eMule AI automatically monitors connectivity and reconnects when the link comes back.
Please enter URL address in "https://www.google.com" format, otherwise you can break this functionality.
When enabled, eMule AI can use NAT traversal transports for peers behind NATs or firewalls.
Choose how NAT traversal transports are selected. Prefer QUIC uses QUIC when both clients support it and falls back to uTP when needed. uTP only disables QUIC completely.
Default mode. Use QUIC whenever both clients support it; otherwise fall back to uTP for uTP only clients.
Force NAT traversal to use uTP only. QUIC is disabled completely and will never be attempted.
Accept incoming uTP by IP-only when the observed remote port is within ±N of the expected port. Helps sequential port-mapping NATs. Larger windows increase wrong-match risk; smaller windows reduce success on remapping NATs. Typical: 64–128. Set 0 to disable IP-only fallback.
Maximum +-N port sweep used for NAT-T hole punching around the expected remote port. Smaller values reduce stray probes and wrong matches; larger values improve success on remapping NATs but increase background traffic. Typical: 8-32. Set 0 to disable sweep behavior.
Minimum random delay between uTP hole punch service bursts. Only used when uTP NAT traversal is active.
Maximum random delay between uTP hole punch service bursts. Only used when uTP NAT traversal is active.
Sets the number of clients you can serve as Kad buddies. Served Kad buddies are firewalled users who depend on you to relay Kad traffic. Raising this limit allows more firewalled clients to participate in file sharing and strengthens the network, but also increases your resource usage: typically only a few kbit/s of extra bandwidth per buddy, and around 0.3% CPU load per buddy on a modern high-performance CPU. The default minimum is 5, and the maximum is 100.
Sets the number of clients you can serve as eServer buddies. Served eServer buddies are LowID users who depend on you to establish callback connections via an eDonkey server. Raising this limit allows more firewalled clients to participate in file sharing and strengthens the network, but also increases your resource usage: typically only a few bytes/s of extra bandwidth per buddy, and around 0.1% CPU load per buddy on a modern high-performance CPU. The default minimum is 5, and the maximum is 100.
With Intelligent Chunk Selection (ICS) the clients exchange information about which chunk is being downloaded at the moment, then a client can select and download the less downloaded chunk. When"Try to download preview chunks" option is enabled, preview chunks are still prioritized until they are completed.
ICS has three working modes. These modes and upload/download chunk prioritizations are as follows:
RELEASE: complete sources <= 3: rarest>less downloaded>smallest>random
SPREAD: complete sources between 4 and 10: smallest>rarest>less downloaded>random
SHARE: complete sources over 10: smallest>preview (if selected in preferences)>random
This feature enables the client to retry a connection attempt after it failed, instead of directly dropping the source. This is an enhancement to bring the balance back with the dead source list. This way only clients land on deadsource list that are really no longer reachable.
When clients with a dynamic IP addresses disconnects and reconnects, their IP changes. Unless they reask the clients they want to download from, those clients cannot be aware about the IP change. So it's possible that a client wants to upload to you, but still thinks you have the old IP and then it fails. This feature helps reasking all the other clients as fast as possible in order not to create the mentioned situation. This feature doesn't work for LowID clients.
Informs all clients in the waiting queue when your IP changed, so they will not lose us and their queue position.
Usually eMule AI reasks every client we want to download from every 15 minutes (30min for the clients with no needed parts). Clients in the upload queue lose their places in that queue after one hour. By increasing the reask time you may decrease overhead of your own client and also the network. But please note that you should not increase your reask time when you have low ID!
Fine tune upload slot behavior and buffering for users with faster upstream connections. Leave these options disabled if you prefer the classic eMule AI behavior.
Opens or recycles upload slots when the configured upload capacity is not being used efficiently. Target upload clients is the normal target, and Elastic upload slots can allow temporary overflow while upload bandwidth is underused. Keep this disabled on slow or unstable connections.
Sets the normal target number of active upload clients for high-bandwidth mode. This is not a strict maximum while Elastic upload slots is above 0%. Allowed range is 4 to 100.
Allows extra high-bandwidth upload slots above Target upload clients after upload underfill persists. Range is 0 to 100 percent. Set this to 0% if you want the target upload client count to act as the effective admission limit.
Defines how slow an upload slot must be before it can be recycled. For example, 0.33 means the slot is considered slow when it stays below about one third of the target slot speed.
How long a slot must stay below the slow threshold before it can be reassigned to another waiting client.
Gives a newly opened upload slot time to reach its normal speed before slow slot checks begin.
How long a slot may send no payload data before it is treated as inactive and eligible for reassignment.
Prevents the same client from being recycled again too quickly after a slow or inactive upload slot was reassigned.
Gives under-served shared files a controlled queue score bonus when their all time upload ratio is below the selected threshold. This helps rare or newly shared files get more upload chances without changing the eD2K protocol.
Files with an all time uploaded to size ratio below this value can receive the low ratio queue score boost. Lower values make the boost apply only to more strongly under-served files.
Additional queue score added when the low ratio boost applies. Keep this value moderate so file rarity helps the queue without overwhelming credits and waiting time.
Limits how much payload one client may receive during a single upload session. The slot is rotated only when another queued client can use it, so bandwidth is not intentionally left idle.
Do not apply an extra per session transfer limit. Classic eMule AI chunk and time rules remain in control.
Uses a percentage of the requested file size as the per-session payload limit. This scales naturally between small and large shared files.
Percentage of the file size that one client may receive in a single upload session when percentage mode is selected.
Uses a fixed MiB amount as the per session payload limit. Set the value to 0 to keep this mode effectively inactive.
Fixed MiB amount that one client may receive in a single upload session when absolute mode is selected.
Maximum upload session duration in seconds before the client can be rotated out. A value of 0 disables this extra time limit.
Temporarily allows larger per file download buffers when several files are receiving data at the same time, reducing disk flush pressure on fast connections while keeping a global memory budget.
Upon activation, eMule AI periodically scans all downloading files, identifies empty (filled with binary zeros) fake files, incomplete (partially filled with binary zeros) files, DRM-protected media files and subsequently removes them from the download list. Additionally, it detects invalid file extensions and replaces them with valid ones.
No detection will be performed.
Empty (filled with binary zeros) fake files, incomplete (partially filled with binary zeros) files, DRM-protected media files and invalid file extensions will be detected, then logged. But these files won't be deleted/renamed.
Empty (filled with binary zeros) fake files, incomplete (partially filled with binary zeros) files, DRM-protected media files and invalid file extensions will be detected, deleted/renamed and then logged.
Enables detection of Empty/incomplete files.
Enables detection of DRM protected files.
When activated, the download list will be periodically scanned for files with invalid file extensions. Any invalid extensions detected will be automatically replaced with the correct, valid ones.
When enabled, new downloads track the most common source filename. In 'Only log' mode it only logs the suggested rename. In 'Delete/Rename' mode it renames the file. You can override this per file from the download list context menu.
When enabled, Auto rename to majority name is enabled automatically only for downloads added after this option is turned on. Existing downloads keep their current per file setting.
Defines how many source filenames must agree before eMule AI accepts a majority name. For example, 51 requires more than half of the collected source names to point to the same filename.
Defines the minimum number of matching source filenames required before Auto rename to majority name can rename a download. Set 0 to disable this extra minimum.
Uses the Download Inspector workflow to evaluate downloads against configurable date and amount criteria. This feature only runs while Download Inspector is in Log only or Delete mode. Log only reports matches without deleting. Delete removes matched downloads via the existing safe delete flow. Date criteria are combined with OR. Amount criteria are combined with OR. The feature only runs if at least one date criterion and at least one amount criterion are enabled. If that requirement is met, the final decision is (DateGroup) AND (AmountGroup). Otherwise, no scan is performed and no deletion decision is made.
Matches downloads whose added date is older than the configured number of days. If the date is not available, this criterion is treated as not matched.
Matches downloads whose added date is older than the configured number of days. If the date is not available, this criterion is treated as not matched.
Matches downloads whose last seen complete date is older than the configured number of days. If the date is missing, this criterion is treated as not matched.
Matches downloads whose last seen complete date is older than the configured number of days. If the date is missing, this criterion is treated as not matched.
Matches downloads whose last received date is older than the configured number of days. If the date is missing, this criterion is treated as not matched.
Matches downloads whose last received date is older than the configured number of days. If the date is missing, this criterion is treated as not matched.
Matches downloads whose completed amount is below the configured percentage threshold. A threshold of 0 only matches downloads with exactly 0% completed.
Matches downloads whose completed amount is below the configured percentage threshold. A threshold of 0 only matches downloads with exactly 0% completed.
Matches downloads whose completed amount is below the configured MB threshold. A threshold of 0 only matches downloads with exactly 0 MB completed.
Matches downloads whose completed amount is below the configured MB threshold. A threshold of 0 only matches downloads with exactly 0 MB completed.
Only applies to files that are actually deleted by this criteria-based auto-delete feature while Download Inspector is in Delete mode. When enabled, the eD2k link of each deleted file is appended to download_inspector.txt in the config directory. Log only mode does not write this file. The file is intended to be preserved by migration and config backup flows.
If enabled, downloads removed by auto-delete are not added to canceled.met. This behaves like Cancel (Don't Mark As Canceled) for auto-delete removals.
When Download Inspector is activated, it periodically checks all downloading files. This parameter sets the interval (in minutes) between each check. Note that the first check will always occur approximately 1 minute after eMule AI starts or this feature is enabled. After the initial check, the defined period value will be used.
This parameter specifies the threshold value for downloaded data. Files with a downloaded data amount equal to or greater than this value will be inspected.
Fake files are typically completely empty (filled with binary zeros), but they may also contain some garbage data interspersed with large empty sections. Additionally, there might be incomplete files (partially filled with binary zeros) on the network that you might want to cancel downloading. This parameter sets the threshold for the percentage of empty data relative to the total downloaded data. If this threshold is exceeded, the file will be considered fake or incomplete. A threshold value of 100 means that all downloaded data must be empty (binary zeros) for the file to be deemed fake.
This parameter specifies the lower threshold value for file compression percentage. Files with a compression percentage equal to or greater than this value will be inspected.
Bypasses the empty data check for files with a high compression percentage and accepts them as fake files.
Specifies the upper threshold value for file compression percentage. Files with a compression percentage equal to or greater than this value will be accepted as fake files without checking the empty data percentage.
Here you can find tweaks related to search and search list.
Enabling this option groups all known (shared, downloading, downloaded, canceled) files at the bottom of the search result list. If "Spam" and/or "Blacklisted" features are also enabled then all these files will be grouped as a single group. When you sort search list this will be sorted as a group and will still stay at the bottom.
You can disable this option to prevent accidental clicks n close buttons that could lead to closing search tabs.
"Mark As Spam" performs complex heuristic analysis and tries to mark possible similar files. During this process, it generates a value which indicates the probability of being a spam file. This value is compared with the default threshold value 60 and if it is more than or equal to the threshold then the file is marked as spam. Here you can increase the threshold value to minimize the possibility of having files marked as spam wrongly. A value equal or greather than 60 should be defined.
Limits how many results one eD2K search may collect before it is stopped. Use 0 for unlimited results. A lower value can keep very large result sets from slowing the search window, while a higher value may find more files.
Limits how many additional More requests can be sent for one eD2K search. Use 0 for unlimited requests. Lower values reduce server traffic and UI noise, while higher values may collect more results from supporting servers.
Controls how many file search answers a Kad file lookup may collect before it stops early. Higher values can return more sources, but may keep the Kad search active longer and increase network work.
A file search started on the Kad Network stops after a specific time limit or when the number of found files reaches a certain limit. Here, you can adjust this limit for the number of found files.
Controls how long a Kad file lookup may stay active. A longer lifetime can find more sources on slow searches, but also keeps the search running longer.
Controls how long a Kad keyword search may stay active. A longer lifetime can collect more search results, but also keeps the search running longer and may increase Kad traffic.
Here you can find tweaks related to E-Server.
When connection fails and there is no more server to try in the server list, start again from the first entry. Default: True
If a static server in the server list goes over the max number of retries it is disabled instead of being removed until the restart of eMule AI
Every server connection triggers saving all part files. When downloading hundreds of files, this takes ""A LOT"" of time and causes freezes. Enabling this option prevents this from occuring. Default: True
Options for file history visibility and NTFS file time daylight saving adjustments and tolerance handling.
Shared files and file history lists can include or exclude downloading files. You can switch this here.
File history list can include or exclude shared files. You can switch this here.
On NTFS file systems, file timestamps may appear to shift by one hour if Daylight Saving Time (DST) is active. Enabling this option adjusts file times accordingly, which may help match files more accurately across DST changes.
WARNING: This adjustment may cause unnecessary rehashing of files when DST switches, potentially leading to performance issues, especially with a large number of shared files. To reduce this impact, consider enabling the "Allow DST time tolerance".
Enabling this option allows a 1-hour tolerance when comparing file timestamps, which helps prevent unnecessary rehashing of files caused by Daylight Saving Time (DST) differences. This is especially useful when the "Adjust NTFS daylight file time" option is enabled, as it may shift file times by one hour depending on DST status. If this tolerance is not allowed, identical files may appear different and get rehashed each time DST switches, which can negatively impact performance, especially with large numbers of shared files.
Options for duplicate history entries, automatic subdirectory sharing, excluded extensions, spreadbar defaults, overshare filtering, selective sharing, Share Only The Need, Powershare and share permissions.
File history list can include or exclude duplicate files. You can switch this here.
This option ensures that all existing and newly created subdirectories are automatically shared when a folder is shared.
Excludes files with the defined extensions when sharing a folder.
Define a comma seperated list for extension that will be excluded when sharing a folder.
Enable collection data for the spreadbar.
This is optimal for better releases but can take some extra memory.
Balance the upload of chunks within a file.
It hides chunks that were uploaded significantly more than other chunks from downloaders. This causes downloaders to request other, less shared, chunks.
There are always some chunks visible.
You can make Hide Overshares stricter by revealing only one chunk to each user, starting with the least uploaded chunks and considering the chunks that were already offered when choosing the next.
Hide Overshares must not be disabled for this to work.
Off by default.
Share Only The Need hides chunks that were already requested by other clients in your queue. eMule AI analyzes your upload queue to see whether all users ask for the same chunks.
If true, eMule AI hides the chunks that at least one user already has and shows only the chunks that nobody has.
There is always one visible.
Powershare settings give absolute priority to a file without the need to unshare other files.
Powershare is off unless it is manually set on an individual file.
Powershare is active unless another mode is manually set on individual files.
Decide whether Powershare is useful depending on the request count and file rarity.
Limit Powershare until the file reaches the configured amount of complete sources.
Amount of virtual complete sources required before limited Powershare stops prioritizing the file.
Use the other priorities of files using Powershare as individual absolute priorities. For example, high priority always goes before normal priority instead of only acting as a factor.
Disabled by default.
Default permission for showing peers a list of your shared files.
All shared files are visible to everyone unless an individual file overrides this setting.
Allow only friends to see a list of your files.
All shared files are hidden.
When enabled, Shared Files rows are colored according to the effective Show Permission setting. When disabled, no permission-based coloring is applied.
Most of the other clients in the eD2k network prefer sharing with clients of their own kind. There are also some communities (which uses addons in the user names) prioritize uploading to their members. In order not to be in a disadvantage, you can activate these features which makes eMule AI pretend to be a client of the same kind of the connected clients and/or a member of the community that connected clients are belong to.
eMule AI will pretend to be a MLDonkey client when enabled.
eMule AI will pretend to be a eDonkey client when enabled.
eMule AI will pretend to be a eDonkey Hybrid client when enabled.
eMule AI will pretend to be a Shareaza client when enabled.
eMule AI will pretend to be a Lphant client when enabled.
eMule AI will pretend to be a member of the community that connected clients are belong to by copying their community tags. This feature implements a smart learning mechanism: Analyzes all tags used in the user names of the all connected clients and if it encounters a tag (on different clients) nth time, where n is equal or more the defined treshold value during an eMule AI execution session, this tag will be saved in "preferences.ini" and started to be emulated. Any tags which is not saved will be ignored and dropped when sending user name.
When eMule AI encounters a tag (on different clients) nth time, where n is equal or more than the defined treshold value during an eMule AI execution session, this tag will be saved in "preferences.ini" and started to be emulated. Any tags which is not saved will be ignored and dropped when sending user name. Default value is 3 and minimum allowed value is 2 since 1 can cause other mods ban/punish our client by sending random tags to see if we copy them.
Enabling this option logs all emulation events to the verbose log output. You also need to activate verbose log option under Options-->Extended-->Verbose to see emulation event log.
What credit system to use. A credit system is required in Morph. The priority of the peers in your waiting queue is based on the credit system.
Start:1 min:1 max:10
Ratio1=Uploaded Total x 2 / Downloaded Total
Ratio2=SQRT(Uploaded Total+2 )
Both ratios are compared and the lower value is used as modifier. Some boundary conditions also exist:
Uploaded Total less than 1MB => Modifier=1
Downloaded Total=0 => Modifier=10
Start:1 min:0.1 max:100
Ratio:1:1.5
l-modifier=100*((1-1/(1+exp((3*{MB uploaded to us}^2-{MB downloaded from us}^2)/1000)))^6.6667)
Only clients using the 'SecureHash' are able to get a multiplier of 100. All others will stay at 10. In contrast to the original credit system, credits are evaluated more on differences and not on quotients. Using the original system you have the best credit values shortly after generating a new userhash. With the new credit system you get good credit values faster if you have already uploadeded many MB before (and did not cheat by killing the userhash).
In contrast to its name, it is not based on UL/DL ratio, but on DL, UL values and their differencies.There is no simple and the only formula.
Start:1
Progressively punishes near zero (<= 1 MB) uploaders.
Progressively boosts uploaders who got nothing paid back yet (<= 1 MB in total).
For client who uploaded to me > 1 MB and <=9 MB grants minimal credit DL/9.
For client who uploaded to me > 9 MB minimal credit is 0.7 + sqrt (DL+1)/10.
With balanced ratio credit raises proportional to sqrt (DL+1)
Credit is farther increased/decreased, depending on UL - DL difference.
No limiting values
Start: 3 min: 1 max: 100.
Ratio = 3 * downloaded / uploaded
For new clients (downloaded and uploaded data less than 1MB)
ratio = 3.0 (instead of 1.0) - If you have received more than 1MB from someone but haven't given anything back user gets ratio = 100.0
Small bonus for clients that have given you many MB: If you get 100MB then user receive ratio = 50.0 till you give back 108MB.
Start:100 min:10 max:5000
Ratio=+6 per MB uploaded and -2 for downloaded
bonux +100 if upload 1MB+
low value if uploaded at least 1MB+, ratio = 50
Magic Angel credit system based on the official Creditsstem. The differences are:
it gives Credits for Upload more than 1.65MB (official 1.00MB)
the lowest modifier is 0.1 (official 1.0)
the highest modifier is 50.0 (official 10.0)
Magic Angel+ credit system based on Magic Angel Creditssytem, but increases the modifier a bit, if the client uploaded to us more than we uploaded to him. If the upload of the client minus his download is
<7MB: modifier+0.3
<15MB: modifier+1.0
<30MB: modifier+2.0
>=30MB: modifier+3.0
Ident Needed --> Ratio=0.75
Ident failed --> Ratio=0.5
Bad client --> Ratio=0
If upload-download>=1GB --> ratio=32
If 0<upload-download<1GB --> ratio=sqrt(upload-download in MB)
Otherwise ratio=1
There are two different credit modifier calculated:
Ratio1=Uploaded Totalx2.2/Downloaded Total
Ratio2=SQRT(Uploaded Total+2)
Both ratios are compared and the lower value is used as modifier.
Some boundary conditions also exist:
Uploaded Total<1MB (exactly!) => Modifier=1
Downloaded Total=0 => Modifier=10
The modifier may not be lower than 1 or higher than 100
If you are sharing completed files and someone downloads data from one of these from you, they probably cannot give you data back as you have all of the file. So in the TK4 system their credit rating remains unchanged. If you are downloading a file and someone takes data from you from the parts of the file you have they will be subject to the credit system and depending on how much they have given they may have their credit rating reduced. At any time if anyone gives you data they get a credit rating increase. The credit system works basically like this:
Start:10 Allowance:'Mb given'/4 + 1MbIf 'Mb taken' greater than 'Mb given + allowance + 1Mb (they have taken more than they gave + some)
Then if the file is a file we are downloading
If '(Mb taken - allowance)' is between 1 to 81Mb credit multiplier range = 9 to 1 math: 10 - square root('(Mb taken - allowance)' - 'Mb given') if '(Mb taken - allowance)' is greater than 81Mb credit multiplier range = 1 to 0.14 math: 9 / square root('(Mb taken - allowance)' -'Mb given')
If 'Mb given' is greater than 'Mb taken' (they have given more than they have taken)
Credit multiplier range = 10 to 342 math: 10 + log(2.72+('Mb given'-'Mb taken)*4)+('Mb taken/12)
There is no absolute range for multipliers but I have quoted a range of 0.14< to >342 (Assumes around a max. 4Gb difference) with a starting point of 10. Clients with failed SUI will start at 10, but not go up.
It rewards clients which gives you a high download. This clients gets a bonus factor.
On the other side, clients you upload much data and the don't give something back to you will get a penalty for the current emule session.
Formula for positive bonus:
Bonus = (download - upload)/10485760 - 1.0/(download/10485760)
The max score ratio is 10. (like official)
A client can get a negative bonus of 0,1 if you gave him 1 chunk(9,28MB) more this session and also at complete comparision of download/upload without geting something back.
A client can get a negative bonus of 0,2 if you gave him more than 2 chunk (9,28MB) this session and also at complete comparsion of download/upload without geting something back.
Start:1 min:1 max:10
Ratio1 = Uploaded Total x 2 / Downloaded Total
Ratio2=SQRT(Uploaded Total+1)
Both ratios are compared and the lower value is used as modifier. Some boundary conditions also exist:
Uploaded Total=0 => Modifier=1
Downloaded Total=0 => Modifier=10
You can enable and tweak setting related with saving and loading clients here. When this feature is enabled, clients will be reloaded after restarting eMule AI. Enabling this feature also helps protection panel, since all punishments will be restored whenever you restart eMule AI.
Enable this feature
Clients not seen since the defined number of days will be purged from clienthistory.met file.
Enabling this option logs all client history events to the verbose log output. You also need to activate verbose log option under Options-->Extended-->Verbose to see client history event log.
Here, you can enable and tweak setting realated with remote client's shared files.
Default tab name is user name if it isn't empty otherwise user hash. But there're many clients using same user names, so it can be difficult to distinguish their tab names in the search list. If you activate this option, user hash will always be used as the tab name.
If this option is enabled and a note has been added to the client, it will be appended in square brackets at the end of the tab name.
This is the time interval in minutes between shared files auto query processes executed.
This is maximum number of clients that will be auto queried in the defined period.
This is the minimum time in minutes between consequent auto queries for the same client. Minimum value that can be defined is 30 minutes.
When enabled, the auto query will be activated for any clients which hasn't queried yet and has downloaded from us more than the specified threshold value.
Auto query will be activated any clients which hasn't queried yet and has downloaded from us more than this amount (in MB) of data.
When enabled, auto query will be activated for any clients which hasn't queried yet and has uploaded to us more than the specified threshold value.
Auto query will be activated any clients which hasn't queried yet and has uploaded to us more than this amount (in MB) of data.
You can enable and tune saving and loading sources here. When this feature is enabled, sources will be reloaded after restarting eMule AI.
Enable this feature
You can limit number of sources here. This can be helpful to reduce saving and loading time and usage of system resources.
You can adjust the expiration day count for sources here. Expiration helps minimize saving and loading times and reduces system resource usage.
You can tune .met files expiration behaviour with these options.
Old entries in the known.met file will never be purged.
Parts of old entries in the known.met file will be purged. This is the way the official does it.
Old entries in the known.met file will be purged completly.
Clean up .met files after file has not been seen for the defined days. Saves diskspace. Applies for known*.met. Never set a very low value (less than 30) here.
Enable if AICH hash of unshared files should be deleted immediately. Default: disabled
Clients not seen since the defined number of days will be purged from clients.met file. Minimum value that can be defined here is 150 days.
These settings allow you to backup your eMule AI configuration data. Here, you can enable and adjust the backup settings.
Enable automatic backup when eMule AI closes.
Enable automatic backup at defined periods.
This is the interval, in hours, at which a new backup will be created.
This is the maximum number of backups. When this limit is reached, the oldest backup folder will be deleted to create a new backup.
This option recursively compresses the backup folder. It is only supported by the NTFS file system.
Minor Preferences for advanced tweaking. Also supported in official by editing preferences.ini
Additional settings for the MiniMule feature.
The MiniMule window will usually remain visible until closed. Setting this value to 0 will automatically close the window when the cursor is moved out of the window. Similar to a tooltip. (default=1)
Controls the transparency of the MiniMule window. 0 and 100 are opaque 1 is translucent. (default=0), might require restart.
How to display some items in the download window
eMule AI remembers which of the main windows (Kademlia, Server, Transfer, was open, when before application exit and restores it. (default = unchecked)
Remember which of the log panes (Log, Verbose) has been used the last session.
Changes the look of the statistic graph window and some other minor styles stuff. (might require restart)
Adds right-to-left reading support to eMule AI. Useful for languages like Hebrew etc.
Enabling this option sets font of the active download as bold in download list.
Defines the maximum number of lines saved in the history of the irc messages. After this value is reached the oldest message is purged on new entry (FIFO - First In First Out). default =100
Maximum open chat sessions
Date&Time format
Date&Time format for lists
Show an vertical line in the statics window every hour. Resize statistics window to see effect (default: enabled)
Option to disable toolbar options like i.e.: Added chevron support: If toolbar is too small to display all buttons a chevron ('>>' icon) is shown at the very right side which opens a menu with the remaining button commands. Tooltips for truncated button titles and in 'No Labels' mode of toolbar. Buttons can be moved and/or removed from the toolbar by holding down Ctrl+Alt key while draging a button with the mouse. Button sizes more similar to Windows Explorer. (default: checked)
Flash the system tray icon when a new message is received. (default=on)
In the shared files windows, Show Copy link to clipboard menu option instead of the 'ed2k links....' context menu. (default = off)
The information displayed in the queue list will be updated not as often which will save some CPU time.
eMule AI may freeze briefly when the window is resized. This is caused by redrawing the graphs with saved data. If this option is checked, the graphs will not be recreated and start anew.
Windows 7 allows overlays for the taskbar icon. eMule AI can utilize this to indicate active up- and downloads. Default: disabled
Additional settings for customizing the logging.
Defines the amount of system memory eMule AI uses to display the logs. For Win XP a default 128 kB reserved. The logs are rotating, i.e. if the buffer is filled up to its limit an old message is purged when a new one arrives (FIFO).
Changes the data format of the log files from unicode to UTF8 when set to 1.
Date&Time format for log
When this key is enabled eMule AI will automatically save a crash dump should eMule AI unexpectedly crash. This can help the developers to track down the problem. The dump is saved in eMule AI's installation folder, called 'eMule-version-number-Modversion.dmp'. Default=unchecked (disabled)
Allows the execution of multiple eMule AI instances. Note: For two eMule AIs running on the same machine they must be installed to different folders, must have different temp folders and also the ports used by each eMule AI have to be unique. It is not recommended to run two eMule AIs concurrently apart from testing purposes. default:0
Name of the certificate for the encryption of email notifications with S/MIME. The certificate will be looked up in the local windows address book.
This function should be only activated when using VLC - VideoLAN Client to preview files. It allows to preview multimedia files when at least 16 kB data for .mpeg files or 256 kB data for all other multimedia content (.avi, CD-Images with multimedia content) has been downloaded. Using this option the 'Create backup for preview' setting has to be deactivated in the 'Video Player' dialog. Note that vlc version 8.5 and higher seem to have some problems with broken indexes.
When this option is disabled (=0), eMule AI will use a new preview method which is faster and uses less disk space in the temporary folder.
If this setting is enabled, double clicking on a row in the download list will invoke the preview command for the related file. If 'Download list double-click to expand' is turned on in 'Preferences -> Display', clicking on the first column invokes the preview command and clicking on the other columns expands the row.
Security Zones as per Internet Explorer Security settings. Defines which features are allowed in HTML code. Relevant for the MiniMule window. Allowed=Untrusted|Internet|Intranet|Trusted|LocalMachine
The default program for editing the Webservices and IPFilter.dat may be entered here. Provide the complete path to the new executable if not using the default notepad.exe.
Normally the local port for asking servers for sources or searching files on them is randomly chosen . If eMule AI runs on a machine in a more complex network, especially when routers are used, it may be necessary to fix this port to a defined value or to deactivate it at all. Disabling this port means that eMule AI will no longer be able to update the status of the servers in its server list and will also not be able to receive sources from them nor do a global search. Exempt from this is the server eMule AI is actually connected to. Usually the default setting should not be changed. Possible values for ServerUDPPort are: 0:Disabled / 1-65534: Fixed local UDP port / 65535:Random port (default)
This setting affects the Delete From Disk function in the right click menu of the Files lists. If disabled then files will be permanently deleted from disk without moving them to the recycle bin.
Option to make eMule AI use highres scheduling (better precision for sleeps and some timers). Currently the upload throttler takes advantage of it if enabled, which may give smoother upload performance and smoother pings while using eMule AI.
Trusts every AICH hashset during the current session. Not recommended except debugging.
Intelligent Corruption Handling (ICH) helps recovering corrupted data downloaded from other clients.
Do not create a new eMule AI_secure account, but restrict the rights of the current account. Might be an alternative for securemode if you are running in a limited account.
When enabled this will use the same server list sorting as shown in GUI when trying to connect to a server.
Amount in MB that are allowed to be downloaded over http from the webserver (Default 5,0=all)
IP addresses that are allowed to connect to the webserver, seperated by ';'
Debug search detail level (Default=0)
Maximum number of random padding bytes added to obfuscated TCP handshakes (default 128). Lower values reduce overhead, higher values increase packet size variation but also add more overhead.
Normally all file data is compressed before transfering, except for rar, zip and some other archive formats This options disables compressing avi files like rar files For high speed upload links this increases performance, but for some avi file some possible compression is lost. (default= false). Note that Morph has the option to disable/decrease compression for all files.
If enabled eMule AI will automatically rearrange search keywords to avoid limitations by Kad. Default: enabled
Beep on errors
Accept messages from valid clients only
Usually shared folders that are unavailable are being removed. Enabling this option will keep those folders if they are on fixed drives so they can be loaded again when they are available again. Default: disabled
No download validation will be performed.
Show a message box and ask user to approve or reject the download when a known file with same/similar name is found.
This option will automatically reject any download request having an entry in the file history with the same file name.
This option will automatically accept any download request having an entry in the file history with the same/similar file name if new file size is larger than the old one by the defined percentage. 0 means accept all; 10 means files %10 or more larger will be accepted.
When "Accept same/similar name conditionally" is selected, you can define a percentage here: 0 means accept all; 10 means files %10 or more larger will be accepted.
Enabling this option will Reject files with a corresponding entry in the canceled history having similar hash or name.
Enabling this option will Reject files with same hash even if file names are different.
Enabling this option will reject files that are manually blacklisted or have blacklisted names. Please note that this option requires the 'Enable manual blacklist' setting to be enabled in the Blacklist Panel.
Enabling this option will always perform case insensitive file name comparison.
Enabling this option will ignore file extension during file name comparison.
Enabling this option will ignore tags enclosed in () [] {} <> during file name comparison.
Enabling this option will include numeric-only tags enclosed in () [] {} <> during file name comparison even though "Ignore tags" is enabled. For example tags like (25) or [25] or {25} or <25> will not be ignored.
Enabling this option will ignore any non-alphanumeric characters during file name comparison.
Enabling this option will detect and strip corrupted character sequences (Mojibake artifacts) from file names prior to all validation checks.
Anything shorter than this file name length will discard comparison and accept file. Results of name processing options "Ignore file extension", "Ignore tags" or "Ignore non-alphanumeric" will be reverted if previous step satisfies this condition. Defined value should be greater than 4.
When both files have a media length found by local MediaInfo analysis, candidates outside the allowed difference are ignored. This check is skipped when either value is unavailable or comes from remote metadata.
Sets the largest allowed difference between media lengths found by local analysis.
When enabled, Download Validator detects date text with hour/minute in the processed file name after the existing name processing options. It uses the exact detected date/time text to find already known or downloading files and treats a date/time text match as a name match without comparing the remaining file name; if the size percentage condition is enabled, that condition still has to pass. Alternative date/time formats are not considered equivalent. Optional sub-options can also require seconds or include a numeric value that immediately follows the detected date/time text; when that numeric option is enabled, the optional separator before the numeric value must match too.
Only detect date text whose resolved year is within the selected four digit range.
Four digit first year used by the date/time detector when year range limiting is enabled.
Four digit last year used by the date/time detector when year range limiting is enabled.
Require seconds to be present and equal in the detected date/time text. When disabled, only year/month/day and hour/minute are compared.
When enabled, numeric values directly after the detected date/time text are included in the comparison key. The values may follow directly or be linked with one separator from space, - _ , . ; : + or %. Separator characters are part of the key, repeated separators stop the scan, and a different or missing trailing value prevents a date/time only match.
Loads regular expression rules from downloadvalidator.conf and checks them against the original full file name. Lines beginning with # after optional spaces are comments. Each active rule must match once and return one or more capture groups with values. Files are treated as having the same name when the same rule returns the same values. Conflicting results from different rules are ignored. Use Reload Configuration after editing the file outside eMule AI. These rules are checked only when exact and date/time checks find no match.
Builds an index from file name character groups to find possible known file matches after exact and regex checks. This option is disabled by default.
Only fuzzy candidates with this score or higher can be treated as similar. The default value of 100 accepts only normalized names that are exactly the same.
Fuzzy matches below this score are not shown under Possible Known File Matches. Trusted same hash, exact name, date and time, and regular expression matches remain at 100 percent.
Require at least this many different whole words to be shared before token coverage can qualify two names as possible matches.
Require shared whole words to cover at least this percentage of the name with fewer unique words.
Use whole name edit comparison only when the shorter normalized name reaches this percentage of the longer name length.
Accept whole name edit evidence only when the normalized names reach at least this similarity percentage.
Minimum number of letters required in a file group label used for ID based matching. Higher values reduce short label false matches.
Minimum number of digits required in the main file ID. Parts after a hyphen may contain fewer digits.
Skip "Potentially Incomplete File" confirmation messagebox and download incomplete files.
If you enable this option, all rejected files will be automatically added to the manual blacklist, unless they are already known files (shared, downloading, downloaded, or canceled). Please note that this option requires the 'Enable manual blacklist' setting to be enabled in the Blacklist Panel.
If you enable this option, once a search is completed, a background process will start to check all search results. Any rejected files will be automatically added to the manual blacklist if they are not already known (shared, downloading, downloaded, canceled). Please note that this option requires the ‘Enable manual blacklist’ option to be checked in the Blacklist Panel.
Enter one regular expression per line to recognize alternate names for the same content. Use capture groups for the parts that must match. Lines beginning with # are comments.
Check the current rule text for syntax errors without saving or activating it. If a rule is invalid, the affected line is selected.
Reload the rule file from disk and replace the text in the editor. You are asked for confirmation when unsaved changes would be discarded.
Show whether shield.conf loaded successfully and how many protection definitions are active. If loading failed, its rules are not being used.
Reload shield.conf from disk and apply its current protection definitions without restarting eMule.
You can set general protection panel system options here.
You can set punishment cancellations timing options here.
Protection Panel scans all punished clients periodically and cancel bans and punishments if they are expired. You can change this period value here. This value is in minutes and should be between 2-20.
Time a client is banned in hours. Please note that all punishments will be reset whenever you restart eMule AI unless you've already enabled Client History feature. Valid value range is 1-720.
Time a client is upload banned or score reduced in hours. Please note that all punishments will be reset whenever you restart eMule AI unless you've already enabled Client History feature. Valid value range is 1-720.
When enabled, clients in the friend list will not be banned by the Protection Panel System.
If a client gets a slot for a particular file, then it can requests and download another file (which is called as file hot swapping) with a lower priority. This is a security leak that a leecher mod could exploit and also known as "Maella Backdoor". When this option enabled, file hot swapping with a lower priority will not be allowed, but file hot swapping with the same or a higher priority will still be allowed.
Generates a user name including a warning text tag and send this to the bad clients.
You can define your own custom text message used in the user name to inform bad clients here. Default text messages will be used if you leave this empty
Don't increase client score before it uploads more than the limit defined below. "Try to transfer full chunks to all uploads" need to be enabled to utilize this option.
Limit before a client gets credits in KB. A value between 1000-2800 should be defined.
Not every client detected by the Protection Panel Systems is really bad. Activating this feature prevents clients from being banned by comparing the defined limit with the upload amount or upload-download difference of a client. Score reduced clients may regain a trustworthy status.
You can set the limit (in KB) as the minimum upload amount or upload-download difference of a client that will activate punishment prevention.
Don't ban the clients who uploaded more than the defined limit.
Don't ban the clients who has a "upload-download" difference more than the defined limit, until the difference reaches the limit.
Don't ban the clients who has a "upload-download" difference more than the defined limit, until the difference reaches 0 (If session ends before 0 is reached, then the client will have to reach the limit again in the next session).
Shield is a customizable dynamic leecher protection module that identifies and punishes leecher clients. It represents an enhanced version of eMule AI XTreme mod's DLP. Unlike DLP, which relies on DLL files, Shield examines client fingerprints defined in the shield.conf configuration file, allowing easy modification by the user.
Checks mod names defined in shield.conf to detect leecher or fake mods, bad communities or other kinds of bad clients.
Checks user names defined in shield.conf to detect leecher or fake mods, bad communities or other kinds of bad clients.
Checks user hashes defined in shield.conf to detect leecher or fake mods, bad communities or other kinds of bad clients.
Hard leechers are defined in shield.conf. They leech community with extremely bad upload/download ratios or not uploading at all. Choose an action to be taken against detected clients.
Soft leechers are defined in shield.conf. Even though they upload fairly, their client softwares have bad features and not approved by the community. Choose an action to be taken against detected clients.
Ban both client's IP address and user hash.
Ban client's user hash.
Don't upload to these clients.
Reduce client score by multiplying 0.1.
Reduce client score by multiplying 0.2.
Reduce client score by multiplying 0.3.
Reduce client score by multiplying 0.4.
Reduce client score by multiplying 0.5.
Reduce client score by multiplying 0.6.
Reduce client score by multiplying 0.7.
Reduce client score by multiplying 0.8.
Reduce client score by multiplying 0.9.
No punishment action will be taken against the clients of this type.
Detect bad Kad nodes by analyzing their behaviour and ban their IP addresses. Banning these nodes slows down their spreading inside Kad network.
Detect wrong packages sent by clients causing processing errors and ban their IP addresses. This could either be due to a malfunctioning/misconfigured client or a malicious attempt resembling a DoS attack.
Detect Anti-P2P bots like MediaDefender and others by checking their user hashes.
Choose an action to be taken against detected clients.
Detect unknown hello and info tags sent by the clients. These are usually unofficial modification to the original code resulting broken functionality and/or buggy messaging. Bad clients like Applejuice and Darkmule may cause this.
Choose an action to be taken against detected clients.
Detect wrong hello and info tags sent by the clients. These are usually unofficial modification to the original code, may or may not be a sign of a leecher client. So be carefull about the punishment.
Choose an action to be taken against detected clients.
Detect clients using our client's hash information and spending our credit information by imitate our client.
Choose an action to be taken against detected clients.
Detect clients faking our client's Mod ID.
Choose an action to be taken against detected clients.
Detect clients using our client's user name.
Choose an action to be taken against detected clients.
Detect eMule AI clients using empty user name
Detect clients changing their Mod ID very often. Using another eMule AI Mod with same configuration folder can cause this and this is generally OK. But changing Mod ID very often might be a sign of the leecher mods.
This is the time interval in minutes that will be inspected to decide if Mod ID change treshold is reached. A value between 30-1440 should be defined.
If this amount of Mod ID changes is reached inside defined time interval, the client will be punished. A value between 1-24 should be defined.
Choose an action to be taken against detected clients.
Detect clients without a mod name (acts as official client) changing their user names very often. User name changes are generally OK, except a client does this very often. This might be a sing of bad purposes.
This is the time interval in minutes between user name change checks which will be used to decide if the user name change threshold is reached. A value between 30 and 1440 can be defined.
If this amount of user name changes is reached inside defined time interval, the client will be punished. A value between 1-24 should be defined.
Choose an action to be taken against detected clients.
Detect clients causing continuous TCP errors. While these errors are not uncommon, the frequency can be alarming with certain clients. This could either be due to a malfunctioning/misconfigured client or a malicious attempt resembling a DoS attack.
This is the time interval, in minutes, that will be inspected to determine if the TCP error threshold has been reached for a specific client. A value between 30 and 1440 can be defined.
If the number of TCP errors reaches the defined threshold within the specified time interval, the client will be punished. The minimum value can be 7.
Choose an action to be taken against detected clients.
Detect bad community tags in client user names.
Choose an action to be taken against detected clients.
Detect clients faking client version.
Choose an action to be taken against detected clients.
Detect eMule AI clients with a mod name containing hexadecimal characters.
Choose an action to be taken against detected clients.
Detect eMule AI mods sending empty mod name.
Choose an action to be taken against detected clients.
Detect clients spamming chat messages.
Choose an action to be taken against detected clients.
Detect bad eMCrypt mods.
Choose an action to be taken against detected clients.
Detect clients exploting source exchange.
Choose an action to be taken against detected clients.
Detect clients that say they don't have a file but request it anyway.
Choose an action to be taken against detected clients.
Ban clients that failed many upload attempts in a row.
Choose an action to be taken against detected clients.
Detects clients that repeatedly abuse upload requests, queue entries or upload slots by reconnecting, dropping, rotating hashes or failing to request blocks after activation.
Detects clients that enter an upload slot and repeatedly fail to request blocks after activation.
Detects clients that repeatedly request the same shared file and re-enter the queue in a short time.
Prevents repeat offenders from inflating shared file request counters with rapid duplicate requests.
Tracks multiple user hashes from the same IP when they repeat the same upload request abuse pattern.
Disconnects clients after hello if their user hash is already banned.
Choose an action to be taken against detected clients.
Detect clients that ask too often for a file.
Define an average time limit in minutes between file reasks before banned. Defined value should be between 5-15.
Define a counter limit for file reasks before banned. Defined value should be between 3-10.
If enabled, logs the reask times of aggressive clients to the verbose log. This can be useful for debugging and identifying problematic clients.
Choose an action to be taken against detected clients.
Punish clients without a Secure User Identification.
Explanation of SUI:Clients in the network are identified by a unique value called user hash which is used to grant earned credits with other users. eMule AI can use an asymmetric encryption to avoid exploiting or manipulating other users hash values and this is called Secure User Identification. Secure User Identification can be turned on in Preferences -> Security and recommended to use it.
Punish MLDonkey clients without a Secure User Identification.
Punish eDonkey clients without a Secure User Identification.
Punish eDonkey Hybrid clients without a Secure User Identification.
Punish Shareaza clients without a Secure User Identification.
Punish Lphant clients without a Secure User Identification.
Punish aMule clients without a Secure User Identification.
Punish eMule AI clients without a Secure User Identification.
Choose an action to be taken against detected clients.
These checks strengthen protections included with eMule against corrupted data, hash manipulation, automated file scanning, rank flooding and excessive Kad requests.
Ban clients sending corrupted data.
Ban clients changing their hash very often. This is usually a sign of the leecher mods.
Ban clients that asked a few times in a row for files we don't have.
Ban clients that send unrequested rank informations.
Ban clients flooding Kad by sending excessive amount of packages.
This is a multiplier of the maximum number of allowed packages sent in a minute. Defined value should be between 1-4. 1 means a client will be banned if it sends more packages than the allowed number; 2 means a client will be banned if it sends more packages than two times of the allowed number; and so on. Default value is 4.
Choose when search result blacklist rules are applied, whether matches are logged and how automatic matches interact with the manual blacklist.
Enabling automatic blacklist feature will mark files in search results as blacklisted when they contain the defined words.
Enabling this will add "Mark As Blacklisted" menu item to the context menu of the search list. When you mark a file on the search list with "Mark As Blacklisted", it will be listed at the bottom inside spam group as spam file with a distinctive text in "Known" column. This file actually will not be marked as a spam and there are some differences between these two: "Mark As Blacklisted" only marks a specific file with its hash value, while "Mark As Spam" performs more complex heuristic analysis and tries to mark possible similar files too. The performance cost of "Mark As Blacklisted" is lower than "Mark As Spam" and there is also no risk for having a file marked as spam wrongly with this option. Both options can still be useful in different situations.
Enabling this option logs all file name matches of automatic blacklist definitions to the verbose log output. You also need to activate verbose log option under Options-->Extended-->Verbose to see automatic blacklist log.
Remove auto-blacklisted files from manual blacklist
Edit the rules that identify unwanted search results. Validate the definitions before applying them so a syntax error does not replace the active rules.
Edit the rules used to identify unwanted search results. Lines that start with # are comments. Open the Help tab for the full rule syntax and examples.
Check the current rule text for syntax errors without saving or activating it. If a rule is invalid, the affected line is selected.
Reload the rule file from disk and replace the text in the editor. You are asked for confirmation when unsaved changes would be discarded.
Read the supported blacklist rule syntax, operators and examples. Use this reference when creating or troubleshooting custom rules.
For current builds, packaged downloads, and versioned release notes, visit the eMule AI Releases page on GitHub. After downloading the release archive from GitHub, unzip it to any directory of your choice, then run eMule AI from that folder.
For Docker deployments, download eMuleAI_Docker_Package.zip from the eMule AI Releases page. The package contains launchers for Windows, Linux, and macOS plus a shared configuration file. The launchers obtain the official emuleai/emuleai image from Docker Hub and start it with persistent storage and the selected display mode. See the Docker Guide below for the complete end-user procedure.
Windows 10 and Windows 11 are supported natively and can also run the Docker image. Linux and macOS are supported through Docker. See the Docker Guide for prerequisites, launcher instructions, display modes, persistent storage, and networking.
The official eMule AI Docker image runs the Windows x64 application under Wine in a Linux amd64 container. The release package provides launchers for Windows, Linux, and macOS. The launchers download or update the image, create the container, mount persistent data, publish the selected ports, wait for the display service, and open the user interface.
eMuleAI_Docker_Package.zip from the eMule AI Releases page. The official image is emuleai/emuleai on Docker Hub.
Windows and macOS users should install Docker Desktop and start it before running the launcher. Linux users need a working Docker Engine and Docker CLI. Docker Desktop for Linux can also be used.
Install an Xpra client from the official Xpra download page when using DISPLAY_MODE=xpra. Xpra is not required when using DISPLAY_MODE=novnc.
Installer selection, platform requirements, permissions, upgrades, and application-specific troubleshooting may change over time. Please refer to the Docker and Xpra websites for the current details of those applications.
eMuleAI_Docker_Package.zip from the selected release.eMuleAI.conf
Windows.bat
Linux.sh
macOS.commandeMuleAI.confOpen eMuleAI.conf in a plain-text editor. It uses KEY=value syntax with no spaces around the equals sign. Lines beginning with # are comments. All launchers read the same file.
The default configuration is suitable for a normal local Xpra installation. Review at least the following settings before the first run:
IMAGE_SOURCE=registry
REGISTRY_IMAGE=emuleai/emuleai
REGISTRY_TAG=latest
DISPLAY_MODE=xpra
XPRA_PASSWORD=emuleai
RESTART_POLICY=no
DATA_DIR_WINDOWS={LOCALAPPDATA}\eMuleAI
DATA_DIR_LINUX=$HOME/.local/share/eMuleAI
DATA_DIR_MACOS=$HOME/Library/Application Support/eMuleAI
EMULE_TCP_PORT=4662
EMULE_UDP_PORT=4672
XPRA_CANVAS_WIDTH=7680
XPRA_CANVAS_HEIGHT=4320IMAGE_SOURCE=registry for normal release use.DISPLAY_MODE=xpra for a separate desktop window or DISPLAY_MODE=novnc for browser access.DATA_DIR_* path with enough free space. This directory preserves configuration, downloads, temporary files, logs, the Wine prefix, and Xpra data.XPRA_CANVAS_WIDTH and XPRA_CANVAS_HEIGHT at least as large as the complete host desktop. The defaults cover common multi-monitor and ultrawide layouts.XPRA_PASSWORD when display access will not be limited to the local computer.Start Docker Desktop. Then double-click Windows.bat or run it from Command Prompt:
Windows.batOpen a terminal in the extracted directory. Restore execute permission if necessary, then run:
chmod +x Linux.sh
./Linux.shStart Docker Desktop. Open Terminal in the extracted directory. Restore execute permission if necessary, then run:
chmod +x macOS.command
./macOS.commandeMuleAI.conf.IMAGE_SOURCE=registry, checks REGISTRY_IMAGE:REGISTRY_TAG on Docker Hub. If the local image is current, Docker reuses the existing layers. If a newer image exists, Docker downloads the changed layers and updates the local tag.CONTAINER_NAME, then creates a replacement using the selected image. Persistent data is retained because it is stored outside the container.AUTO_ATTACH_XPRA=yes, starts the installed Xpra client automatically.CLEAN_OLD_IMAGES=yes, removes older image tags and unused image IDs only after the replacement container starts successfully.docker pull command is needed in registry mode.
With DISPLAY_MODE=xpra, the launcher connects to tcp://127.0.0.1:14500 by default and opens eMule AI as a separate desktop window. The port, username, password, client path, Windows backend, and scaling behavior are configurable.
With DISPLAY_MODE=novnc, open the URL printed by the launcher. With default settings it is:
http://127.0.0.1:6080/vnc.html?autoconnect=1&resize=scaleRESTART_POLICY=no, the container remains stopped.docker stop eMuleAI when the default container name is unchanged.The default eMule ports are 4662/TCP and 4672/UDP. For High ID, the values configured in eMuleAI.conf, the ports configured inside eMule AI, and the ports forwarded by the router should match:
EMULE_TCP_PORT=4662 -> router 4662/TCP
EMULE_UDP_PORT=4672 -> router 4672/UDPThe launcher maps each host port to the same container port. Avoid translating a different host port to the eMule port because eMule announces its configured listening port to peers.
DISPLAY_BIND_ADDRESS=127.0.0.1 keeps Xpra or noVNC local to the computer. Publishing a display service on 0.0.0.0 exposes it to other network interfaces and should only be done with appropriate network restrictions, authentication, and transport protection.
The selected operating-system data directory is mounted as /data in the container. It contains:
config: eMule AI configuration and state files.Incoming: completed downloads.Temp: incomplete downloads and part files.logs: eMule AI logs.wineprefix: persistent Wine environment.xpra: Xpra authentication and related data.Replacing or updating the container does not remove this directory. Back it up according to your normal data-protection policy.
First confirm that Docker is running and that the selected display port is not already used by another application. Useful commands with the default container name are:
docker ps -a --filter name=eMuleAI
docker logs eMuleAI
docker port eMuleAI
docker inspect eMuleAIFor Xpra problems, confirm that the host Xpra client is installed and that XPRA_CLIENT_EXE_WINDOWS, XPRA_CLIENT_EXE_LINUX, or XPRA_CLIENT_EXE_MACOS points to the correct executable when automatic detection does not find it.
For pointer alignment or moved-window click problems on Windows, keep XPRA_CLIENT_BACKEND_WINDOWS=gtk, XPRA_DESKTOP_SCALING=off, and ensure the Xpra canvas is at least as large as the complete Windows desktop.
For noVNC problems, verify the printed browser URL and the configured NOVNC_PORT.
eMuleAI.conf ReferenceThe descriptions below match the shared configuration file included in eMuleAI_Docker_Package.zip.
Some features require user-supplied external files such as server.met, nodes.dat,
ipfilter.dat, or GeoLite2-City.mmdb.
If you already have these files, copy them to the eMule AI config directory and restart the application.
You can also load server.met from Servers > Update server.met from URL, nodes.dat from
Kad > Nodes.dat from URL, and ipfilter.dat or ipfilter.zip from
Options > Security > IP Filter > Update from URL.
GeoLite2-City.mmdb must be downloaded separately from
MaxMind and is
subject to MaxMind's terms. After copying it to the eMule AI config directory,
enable GeoLite2 in Settings and restart the application.
Please use only sources you trust and review their origin, terms, license conditions, and legal status before use.
eMule AI includes a built-in Migration Wizard. For most users, this should be the first and preferred way to restore an older eMule setup. Manual migration is still available for advanced users who want to copy and review files themselves.
Start here first. On first launch, eMule AI can automatically offer to restore an older eMule configuration after language selection. The wizard checks the default Windows profile location used by classic eMule and can also let you choose another folder manually.
Tools > Migration Wizard. When started from the Tools menu, eMule AI schedules the wizard for the next application start, so you should restart the client.config folder and shows a clear summary when the operation finishes, including any files that could not be copied.Manual migration is still supported for advanced users. eMule AI stores its active configuration in a local config folder next to eMuleAI.exe, not in the Windows user profile location used by older eMule builds.
If you prefer to copy files yourself, close eMule AI first, copy only the files you actually want to keep, and then start the client again. If your older setup stores downloads.txt in the main eMule folder instead of inside config, remember to bring that file over as well.
While manual migration is supported, a clean configuration can still be the better choice if you want to start with eMule AI's modern defaults for current hardware, high-speed networks, and newer connection handling.
If you migrate manually, these files are the most valuable ones to keep because they preserve your identity, credits, and network state.
Your reputation and queue ranking with other clients are tied to your unique identification. Without these files, you lose your "earned luck" and will start from zero in other users' queues.
Preserve your social standing and the credits other users have earned with your client.
Ensure a fast start without having to re-bootstrap your connections or re-scan your entire library.
config folder..p2p or .dat files in the config folder at startup, so no extra import step is needed for those files.This is not a Migration Wizard problem, and it is also not the normal result of a manual migration into a separate eMule AI folder. The real risk appears when eMule AI is copied directly into an existing classic eMule installation folder and started there against the same live configuration.
In that situation, files such as known.met and StoredSearches.met can be updated by eMule AI and may no longer remain compatible with the original eMule client if you later switch back to that same folder.
config directory. If you want to test, compare, or keep a fallback copy of classic eMule, work on a backup or a separate folder instead of reusing the original live installation.
clients.met file. Are my credits gone?clients.met file (clients.met.BAK) so you can restore the file by renaming the backup to clients.met.preferences.dat and cryptkey.dat. Should one of these files be lost or deleted, you will no longer be able to prove that you are the rightful owner of the credits.
cryptkey.dat, it is mandatory to delete the preferences.dat file too.An ID is a value calculated from the client's IP address. It is assigned by the server when eMule successfully connects. The purpose of the ID is to indicate whether proper bidirectional communication is possible between two clients.
eMule displays your ID status graphically as colored arrows on the globe icon in the lower-right corner, next to the connected server. After connecting, the ID value itself is also visible in the Server's Log window.
A High ID means the port chosen in your connection settings (default 4662) is open and freely accessible from the internet. Your client is fully reachable by others.
A Low ID means your connection port is blocked or cannot be reached. This is typically caused by firewalls, routers, or proxy servers. Any ID value less than 16,777,216 is considered a Low ID.
High ID = A + 256*B + 256*256*C + 256*256*256*D.If you're stuck with a Low ID despite having correct settings, try connecting to a different server or using the Port Tester in connection options. Low user servers often have more resources to handle Low ID routing than congested ones.
These files are specific to eMule AI and handle its advanced protection, history, and automation features.
This feature combines NAT Traversal (NAT-T) and uTP to create direct, reliable peer connections even when both sides sit behind NAT or firewalls.
In eMule AI it unlocks LowID-to-LowID transfers with UDP hole punching and uTP, improving connectivity and stability on difficult networks.
Because the NAT-T/uTP signaling and relay flow are eMule AI-specific, both peers must be running eMule AI to benefit; the more eMule AI users online, the more often this path is available.
OP_CALLBACKREQUEST, then continue relay after target comes back.[eServerBuddy], OP_ESERVER_RELAY_REQUEST, OP_ESERVER_PEER_INFO, OP_HOLEPUNCH, and uTP.3 and 100 peers (default 3).Classic Kad-driven connectivity matrix where rendezvous and buddies coordinate hole punching.
| # | Kad | TCP | UDP | Buddy | Outcome |
|---|---|---|---|---|---|
| 1 | Open | OpenInbound OK | Open | Not needed | Use plain TCP/UDP; NAT-T not needed. |
| 2 | Open | FirewalledInbound blocked | Open | Not needed | Direct UDP callback works (LowID->HighID). NAT-T not needed for this path. |
| 3 | Open | FirewalledInbound blocked | Open | Started (to enable rendezvous) | NAT-T/uTP possible via buddy-mediated rendezvous/hole-punch. Useful for LowID<->LowID. |
| 4 | Firewalled | OpenInbound OK | Firewalled | Not needed | Plain TCP works; NAT-T not needed. |
| 5 | Firewalled | FirewalledInbound blocked | FirewalledOutbound usually OK | Uploader's buddy | NAT-T/uTP path via rendezvous + hole-punch. Uploader-side buddy is required; downloader buddy is optional. |
| 6 | None | FirewalledInbound blocked | FirewalledVaries | Unavailable | Kad rendezvous is unavailable. Use eServer Buddy path only if requester has buddy on the same server. |
LowID<->LowID relay driven by eMule AI clients. This path does not depend on server-side NAT relay extensions; it uses OP_ESERVER_* signaling plus classic server callback when needed.
| # | Preconditions | Peers | Outcome |
|---|---|---|---|
| 1 | Same eServer + requester has connected eServer buddy | LowID requester <-> LowID target | Buddy runs OP_ESERVER relay handshake and both peers form uTP with hole punching. |
| 2 | Target already connected to buddy | LowID requester <-> LowID target | Buddy forwards OP_ESERVER_PEER_INFO immediately; no server callback round-trip is needed. |
| 3 | Target not connected to buddy | LowID requester <-> LowID target | Buddy first tries classic OP_CALLBACKREQUEST (if server callback is available). Relay completes only after target connects and ACK is received; otherwise it fails. |
| 4 | Different eServer or no connected buddy | Any LowID pair | eServer Buddy path is unavailable; fallback is Kad rendezvous (if prerequisites exist) or classic callback/TCP paths. |
OP_ESERVER_BUDDY_REQUEST. If buddy changed server, reconnect and wait for a new buddy candidate.Initial public release. See below sections for details.
QUIC is now the primary NAT Traversal data protocol for compatible peers. Direct capability negotiation selects QUIC when available and automatically falls back to uTP for compatibility.
Use the official emuleai/emuleai image with the release launcher package and shared configuration file. Xpra provides a separate desktop window while noVNC provides browser access.
A dedicated settings group controls high-bandwidth slot targets, elastic overflow, slow and inactive slot recycling, low-ratio queue assistance, session transfer and time limits, and adaptive download write buffering.
Bind P2P sockets to an adapter or local address. IP Guard can block network activity until the binding and optional public IP allowlist are verified, including runtime changes.
Progress overlays and list status boxes make parallel startup, safe shutdown, large reloads, and bulk add or delete operations visible without blocking the interface.
sharedcache.dat accelerates large shared-folder scans, remembers verified path metadata, and reduces repeated duplicate checks while remaining disposable and automatically rebuildable.
New eD2K and Kad result, request, and lifetime controls limit oversized searches. Files -> All Shared Files adds a sortable Last Request column.
eMule AI can maintain Windows Firewall listen-port rules and select new listen ports from a configured range at startup.
Notifications can use a tray balloon, Windows toast notification, or the custom eMule AI popup.
Automatic rename decisions can be restricted to new downloads and require both a configurable agreement percentage and minimum vote count.
A dedicated Options page that centralizes safety/anti-leecher controls with clear explanations, timed punishments, and friend exemptions. Choose actions (ban, score reduce, log only) per detection category to keep the swarm clean with minimal false positives.
A full dark modern theme for windows, tabs, menus, list headers, and progress bars. It preserves readability, supports hover/pressed states, and can be toggled without restarting.
Provides a smoother experience in heavy lists (Files, Downloads, Search, Clients) with lower memory usage and faster responsiveness, smooth scrolling and quick updates even with hundreds of thousands of rows.
LowID <-> LowID now works through two distinct NAT-T paths. Kad Buddy path (classic): Kad rendezvous flow where uploader side needs a Kad buddy. eServer Buddy path (new in eMule AI): same-server relay between LowID peers using OP_ESERVER_BUDDY/RELAY packets. This eServer Buddy mechanism is a client-side eMule AI feature and does not require server NAT extension support (no dependency on Lugdunum NAT opcodes). If the target is not connected yet, the buddy can try classic OP_CALLBACKREQUEST and finish relay after target reconnect + ACK. External UDP port discovery, hole-punch retries, QUIC negotiation, and uTP fallback retries improve success under difficult NAT conditions. Extended Source Exchange further improves source discovery and transfer stability with richer peer metadata.
End-to-end IPv6 handling for peers, friends, and sockets alongside IPv4; addresses display and copy correctly across dialogs and tooltips.
A Spam Rating column (0-100), a customizable spam threshold, logging of rating results, and a Blacklist Panel with "recheck now" combine to surface quality results and hide junk.
Lightweight history of clients you met (with retention control) to enable once-per-client automation like first time shared files queries.
Refreshing shared files runs in a worker thread and updates the UI progressively, avoiding stalls while scanning disks.
Background watcher keeps the Shared view in sync with filesystem changes without manual reloads.
When enabled, subfolders under your shared roots are included automatically, both existing and newly created ones.
Before adding a file, checks duplicates/near-duplicates across downloads/history/shared and can warn, reject, or auto-blacklist by your preference.
Language support is embedded, so external language DLL files are not required. Supported language codes (116): en, af, am, ar, ast, az, be, bg, bn, bs, ca, ca-VAL, ceb, co, cs, cy, da, de, el, eo, es, es-AR, et, eu, fa, fi, fr, fy, ga, gd, gl, gu, ha, haw, he, hi, hmw, hr, ht, hu, hy, id, ig, is, it, iw, ja, jw, ka, kk, km, kn, ko, ku, ky, la, lb, lo, lt, lv, mg, mi, mk, ml, mn, mr, ms, mt, my, nb, ne, nl, nn, no, ny, or, pa, pl, ps, pt, pt-BR, ro, ru, rw, sd, si, sk, sl, sm, sn, so, sq, sr, st, su, sv, sw, ta, te, tg, th, tk, tl, tr, tt, ug, uk, ur, uz, vi, xh, yi, yo, zh-CN, zh-TW, zu.
One-click background connectivity test with clear start/stop log lines; helps diagnose DNS/routing hiccups without interrupting transfers.
Optional compatibility/emulation toggles for certain clients/communities to improve interoperability and prioritization.
Quick actions for Save State, Reload Config (no restart), and threaded Backup. Numbered Preview buttons (1-10) launch your chosen preview apps directly; right-click can jump to the next previewable file.
At startup, all *.p2p files in the config folder are merged into the active IP filter, just drop lists in and they apply.
You can limit how many answers a Kad keyword search collects before auto-stopping.
Fast, on-view filters to narrow lists to what you care about (for example, active/paused).
New columns in client lists show Friend status, User Hash, ID Type (High/LowID), and Client Version for quick grouping and sorting.
Shows each peer's current endpoint in a sortable column.
Displays how your credit ratio appears to the remote side for clearer give-and-take diagnostics.
Right-click in Search/Download/Shared lists to cancel downloads; an option lets you cancel without marking as "canceled" in history.
New columns indicate query status, last queried time, and number of files received for a client's shared list.
Attach a short note to any client; optionally show it in the tab title when viewing that client's files.
Copy selected list rows as neat plain text for pasting into notes/spreadsheets.
A command to quickly purge obsolete/blacklisted items from active search tabs.
Turn off type-to-search in heavy lists to avoid accidental scans.
Cancel a download without adding it to the "canceled" history.
Understands ed2k "friend" links from clipboard to add friends quickly.
Temporarily caches discovered sources so paused/resumed files can reuse them without re-querying.
Shows country (and city when known) and a flag icon for peers/servers; can be disabled or limited to country only.
Download-list tooltips include full client software/version lines.
Wide-character support and a "filename_untouched" placeholder in webservices.dat; multi-selection webservice actions.
Built-in live up/down graphs in the Connection page and main window for a lightweight visual of throughput.
Optionally auto-query a client's shared files after first connection, paced by configurable per-period limits.
One-click save of list layouts, column orders, and preferences for easy layout "bookmarks".
If no choice exists yet, uses the app's folder as the base location for a portable-style setup.
Adds a built-in Migration Wizard that can run automatically on first launch after language selection or be scheduled later from Tools > Migration Wizard. It helps restore supported legacy eMule configuration files into eMule AI with directory detection, copy results, and clear follow-up guidance.
Allow local/private addresses even when an IP filter is active, useful on home/lab networks.
File operations (hash/preview/move/backup) work reliably beyond classic 260-char Windows limits by using long-path-safe APIs and formats.
Compact backing stores, batched refresh, selection/scroll restore, and combined/heap sorting make giant lists smooth and responsive.
Only the active tab refreshes immediately; others update counters without full redraws. Virtualized storage accelerates inserts and sorting.
File allocation, metadata flushing, and buffer moves occur inside the writer thread under scoped locks, with periodic disk-space checks, reducing contention and stutters.
Bulk operations coalesce UI work, and large lists use a single combined sort pass for steady ordering during heavy updates.
On public IP changes, Kad reconnects gracefully; peers are informed and sources re-asked to minimize stalls.
Writes sensitive data to a temp file and atomically replaces the original, keeping files intact if something fails mid-write.
Complex Boolean expressions are preserved and appear as tab titles; restored searches match your original filters and fields.
Autocomplete/history for IP:port and URL fields, Enter-to-bootstrap, Delete to remove suggestions; remembers your last choice.
Static IP filter file, separate whitelist, better loader/merger behavior, and corrected save logic (no truncation).
Efficient flag handling and optional long vs. short country names; single-click enable/disable.
Keeps ports/IDs/Kad status current while open, with tidy formatting and consistent refresh.
Menus/tabs get theme-aware drawing, clear submenu indicators, and crisp text rendering.
Cleaner menus, better icons/spacing, and improved multi-level sorting for predictable results.
Gentler retry logic and more realistic counters reduce misleading stats and improve source discovery.
Request bursts are smoothed and reask timers clarified to keep queues fresh without spamming.
Idle clients and stale entries are cleaned more aggressively to keep long sessions lean.
More compact server-publish packets and fewer unnecessary sorts make "publish to server" snappier.
Compatible results are merged while respecting spam heuristics to reduce duplicates and clutter.
Accurate OS detection helps choose safe defaults/paths on modern systems.
Defers redraws during initial load to reduce repaint cycles and speed up UI bring-up.
Opens files/links via GNOME/KDE helpers and translates paths properly under WINE.
Apply web actions to multiple selections at once for faster lookups.
Skips redundant disk writes on reconnects, trimming stutter on large libraries.
Checks the preview-apps map only when opening the menu to avoid periodic disk I/O.
Stores full paths with known files for accurate restoration and display.
Accepts small timestamp shifts due to daylight saving, preventing needless re-hashing.
Collects power-user toggles (Search/Server/File handling, etc.) with concise on-screen explanations.
Quickly step through previewable items from the toolbar without changing selection.
Cleaner header layout reduces accidental clicks while preserving functionality.
Search entries derive type from filename first, falling back to the tag only when necessary for accuracy.
Improves crash-report readability without impacting performance.
Suppresses noisy traces by default and clarifies "missing icon resource" diagnostics for easier skin/theme troubleshooting.
Automatically loops the server list on failures to improve connection reliability.
Waits for full initialization before auto-connect and stops hashing immediately on exit for faster, cleaner shutdowns.
Remembers your preferred search category (Any/Audio/Video, etc.) between sessions.
Avoids accidental command executions from toolbar double-clicks.
Skips redundant saves/re-sorts when a category remains the same.
Connection parameters and related defaults were optimized to better match today's hardware and network conditions, improving balance, responsiveness, and overall transfer behavior.
Guards progress-bar drawing to avoid invalid GDI calls during fast scrolling in the Download list.
Waits properly for internal locks and stops hashing on exit to eliminate "Failed to save HashSet" after hashing.
Loads stored searches early to prevent creation of empty placeholder files at startup.
Skips late completions for just-deleted files to avoid a rare crash during heavy cleanup/shutdown.
Releases sockets correctly when entering the "too many connections" state under load.
Prevents crashes by safely refusing source requests when the current file pointer is missing.
Restores the last used folder for that mode in stored searches.
No longer requires previewability checks; relies on free-space threshold to avoid false negatives.
Fixes a debug-build crash due to a wrong trace format string.
Writes complete addresses without truncation when saving filter files.
Keeps boundaries correct when merging ranges by comparing with computed "end+1".
Logs "Connecting ..." only when a connection is actually pending.
Guards against null dereferences when removing multiple finished items.
Removes empty gaps immediately and clamps boundaries correctly; progress/availability remain accurate.
Percentages use the actual hashed size as denominator to avoid misleading jumps and rare divide-by-zero.
Clears buffers and closes gracefully when media details cannot be read instead of throwing.
Locale-aware, case-insensitive sort by "File Type" behaves predictably with neat sub-sorting.
Deletes custom brushes, detaches/destroys header image lists, and tightens ownership to close small GDI leaks.
Removes invalid anchors safely during layout changes to eliminate resize assertions.
Defensive checks prevent crashes when an error string is unexpectedly null.
Fixes extended multi-packet response handling for better peer compatibility.
Removes unnecessary byte-order conversions that could mislead logging/handling.
Ensures the ED2K clipboard prompt isn't hidden behind the splash screen on startup.
Removes duplicates reliably without orphaning children; child items are also deleted correctly.
Ignores just-deleted downloads when building context-menu state to prevent rare crashes.
Promotes values exceeding 32-bit safely, preventing corrupted metadata and parsing errors.
More conservative logic reduces stalls and prevents missed updates.
ASSERTs are logged to file rather than popping intrusive dialogs in enabled mode, aiding diagnostics without interrupting transfers.
Literal "%" characters in log messages print correctly (no accidental format parsing).
Opens long-path files safely when reading metadata to avoid failures in deep directory trees.
Access the official eMule help resources for setup guidance, feature explanations, and practical usage information.
For official reference material and user guidance, visit the eMule Help page.
For community discussion, user tips, and broader ecosystem conversation, visit the eMule Forum.
The most helpful reports include a clear reproduction path, the exact build you used, and diagnostic files when available.
To report a bug, document a reproducible problem, or suggest an improvement, please use the eMule AI GitHub Issues page.
Some problems can be understood from a normal description and a screenshot, but crashes, unexpected assertions, and difficult network issues usually need stronger diagnostics. The three sections below explain the most useful reporting methods for those cases.
If the issue does not crash the client but still needs deeper analysis, verbose logs are usually the best starting point. They are especially helpful for connection behavior, NAT-T and uTP issues, unexpected state changes, retries, queue logic, and similar problems that are hard to explain with screenshots alone.
logs folder first, so the new logs are easier to identify.Options > Extended.logs folder and attach them to the GitHub issue, or upload them to a file host if they are too large for GitHub.If eMule AI crashes, shows an invalid argument message, or closes unexpectedly, please attach a crash dump whenever possible. A dump usually lets the developers identify the failing function much faster than a screenshot or Event Viewer entry alone.
Options > eMule AI > Advanced Preferences.config folder of your eMule AI installation and attach the newest .dmp file to the GitHub issue.config folder, not necessarily the main program folder.
If no dump is created by the built-in method, you can use ProcDump from Microsoft Sysinternals as a fallback.
procdump.exe -accepteula -ma -e -t -w eMuleAI.exe C:\Dumps
After running that command, start eMule AI normally, reproduce the crash, and then attach the newest dump file from C:\Dumps.
If your problem looks network-related, for example sources stay at Unknown, many sources end up as Too many connections, NAT-T or uTP works only on some networks, or transfers improve after disabling Enable uTP & NAT Traversal (NAT-T), please include a deeper report than a normal bug description.
Enable uTP & NAT Traversal (NAT-T) is disabled.Wireshark is a free network analysis tool. It helps show whether the expected UDP and uTP packets are actually reaching your PC during the failing test. You can download it here: https://www.wireshark.org/download.html.
udp.port == 4672. Replace 4672 with your actual UDP port..pcapng file..pcapng file and attach it to the same GitHub issue together with the matching verbose logs from the same test session.Built with Visual Studio 2026 v18.8.1, using the latest MSVC and MSVC Build Tools available at the time.
Python 3.7 or newer must also be installed and available in PATH, because the build runs srchybrid/translations/ai_translator.py for translation map validation and generated header output.
Building eMule AI requires downloading and compiling additional third-party libraries that are not bundled with the distribution. The required dependencies are listed in the table below. Please note that some entries use official release versions, while others reference a specific commit and branch.
Download and place each library’s source code one directory level above the eMule AI source tree, then compile it. Check the project files for the expected folder names, or adjust the paths to match your layout.
| Library | Version / Commit | Source |
|---|---|---|
| cryptopp | #4775a16@master | github.com/weidai11/cryptopp |
| gmp | 6.3.0 | gmplib.org |
| gnutls | 3.8.13 | gitlab.com/gnutls/gnutls |
| libmaxminddb | 1.13.3 | github.com/maxmind/libmaxminddb |
| libutp | #490874c@post-3.4-transmission | github.com/transmission/libutp |
| mbedtls | 4.0.0 | github.com/Mbed-TLS/mbedtls |
| MediaInfoLib | 26.05 | github.com/MediaArea/MediaInfoLib |
| miniupnpc | 2.3.10 | github.com/miniupnp/miniupnp |
| nettle | 4.0 | gitlab.com/gnutls/nettle |
| ngtcp2 | 1.23.0 | github.com/ngtcp2/ngtcp2 |
| ResizableLib | #bebab50@master | github.com/ppescher/resizablelib |
| ZenLib | 0.4.41 | github.com/MediaArea/ZenLib |
| zlib | 1.3.2 | github.com/madler/zlib |
Many thanks to everyone who contributed to eMule over the years. Without their work, eMule AI would not have been possible.
Developers: Merkur, John aka. Unknown1, Ornis, Bluecow, Tecxx, Pach2, Juanjo, Dirus, Barry, zz, Some Support
Modders: David Xanatos, Stulle, XMan, netfinity, WiZaRd, leuk_he, enkeyDev, SLUGFILLER, SiRoB, khaos, fox88, Enig123, TAHO, Pretender, Mighty Knife, Ottavio84, Dolphin, sFrQlXeRt, evcz, cyrex2001, zz_fly, Slaham, Spike, shadow2004, gomez82, JvA, Pawcio, lovelace, MoNKi, Avi3k, Commander, emulEspaña, Maella, VQB, J.C.Conner
Testers: fangguihua1995, QICKV8, lzk87, Heliotropo, Havokdan, Andrey23, mistressadmin, enone, AnneDane, edelkas, Aokromes, superlent, tchara, OConnell, potes31, lapollarecord, tictoc9, arturx, Sony, Monk, Myxin, Mr Ozon, Daan, Elandal, Frozen_North, kayfam, Khandurian, Masta2002, mrLabr, Nesi-San, SeveredCross, Skynetman