mri-txadmin · mri-fivem-skills

Runs ensure, restart, start or stop on a resource of a local FiveM server through the txAdmin API (no rcon) and shows what fxserver.log recorded.

No rcon: txAdmin runs locally and its panel has an API. The script logs in, sends the command and prints the new lines of fxserver.log.

Set up once: copy config.example.json to config.json (kept out of git) and fill in the URL, the txAdmin login, the server folder (base, where txData lives) and the profile.

Run from the skill folder, Node 18+:

node scripts/tx.mjs ensure rpemotes-reborn
node scripts/tx.mjs restart my_resource
node scripts/tx.mjs stop my_resource
node scripts/tx.mjs ensure job_a job_b job_c   # several at once, on the same login
node scripts/tx.mjs refresh                    # rescan resources (new folder)

Action	What it does
`ensure <res>`	Starts it, or restarts if running. The default to reload a change.
`restart <res>`	Restarts (fails if stopped).
`start <res>` / `stop <res>`	Starts / stops.
`refresh`	Server `refresh`, to see a new resource or a changed `fxmanifest.lua` (new file in `client_scripts` or `server_scripts`): without it `ensure` starts with the old manifest and the client or the server gets `attempt to index a nil value` on the missing file's global. Run it before the `ensure` whenever the manifest file lists changed.
`say <message>`	txAdmin announcement to every player. Only when asked.

Login: txAdmin blocks login for 15 minutes after about 10 attempts in a row (error 429). The script keeps the session in %TEMP%\mri-txadmin-<url>.json and only logs in again when it expires. For several resources, pass them all in one command instead of looping calls. On a 429, wait the 15 minutes: retrying doesn't help.

Flags: --base <folder> and --profile <profile> (defaults in config.json), --url if txAdmin isn't on port 40120, --no-log to skip reading the log.

After the command

Check in the output that the resource started (Started resource <res>) and that no SCRIPT ERROR or Couldn't start resource came up.
Restart only restarts that resource. Whoever depends on it (e.g. an emote menu on top of rpemotes-reborn) keeps the old instance's state and may stop working on the client. Restarted an engine/base resource: also restart those that use it.
Resources that list it in dependencies are stopped with it and don't come back by themselves (e.g. ensure sd-phone takes down sd-tablet and the apps). Read the Stopping resource lines in the log and ensure each one.
Large resource (e.g. sd-phone, 22 MB): on a connected client the dependents may start before it finishes downloading, and it stays mounted without running (No such export in the apps). Check Creating script environments for <res> in the FiveM log; without that line, the player needs to reconnect.
New resource or changed dependencies: no server-only resource (e.g. qbx_vehicles) in dependencies. The client can't find it, refuses the resource and FiveM drops the player ("Could not find dependency ... Killing network"). Check the FiveM log after ensure.
Resource with stream/ or a model data_file: ask first. Remounting it on a connected client crashes the game.
The client side shows in the FiveM log (FiveM.app\logs\CitizenFX_log_*.log, in the FiveM folder), not in fxserver.log.
The script never touches server.cfg.

Errors

login failed: txAdmin down or password changed. Check config.json.
command failed: txAdmin's message comes along (e.g. the resource doesn't exist).
Arbitrary console commands don't go through this API (the Live Console uses a websocket): use the txAdmin console for that.

Server commands through txAdmin

After the command

Errors