anjin-intel

A tiny, stdlib-only agent that tails your EVE Online chat logs and ships intel lines to your anjin server, which alerts you when a hostile is reported near where your character is.

Why it exists: ESI (EVE's API) exposes no in-game chat. Chat lives only as files the EVE client writes to local disk while you're logged in. A server can never pull it — so this agent runs on your PC, tails the logs, and POSTs the lines.

Why it's open + dependency-free: it reads your chat and sends it to a server, so you should be able to verify exactly what it does. It's MIT-licensed, pure Go standard library (trivially auditable, reproducible), and read-only — it tails the log directory and POSTs; it never writes to the game and never touches anything but the channels you explicitly allow. Default is no channels.

Scope: Linux (Steam/Proton, Lutris) and Windows — both install and start the shipper at login. macOS is a planned follow-up: install saves the config there, but you start it yourself.

Get it

Grab the binary for your OS from Releases.

Linux:

curl -fsSL -o anjin-intel \
  https://github.com/polarn/anjin-intel/releases/latest/download/anjin-intel-linux-amd64
chmod +x anjin-intel

Windows (PowerShell):

Invoke-WebRequest -OutFile anjin-intel.exe `
  https://github.com/polarn/anjin-intel/releases/latest/download/anjin-intel-windows-amd64.exe

The binary isn't code-signed (a certificate is hard to justify for a tool with a handful of users), so Windows SmartScreen may warn on first run: More info → Run anyway. The SHA256SUMS + provenance attestation below are the real check.

Each release ships a SHA256SUMS and a SLSA build provenance attestation, so you can verify the binary came from this repo's CI (not a hand-built upload):

gh attestation verify anjin-intel --repo polarn/anjin-intel

Or build from source (pure Go stdlib, no deps):

go build -o anjin-intel ./cmd/anjin-intel

Usage

Linux

Install — registers a systemd user service that runs the shipper at login and copies the binary to ~/.local/bin:

anjin-intel install \
  --server https://anjin.example.net \
  --token  <enrollment-token-from-the-Intel-tab>
  # --logdir is auto-detected (Steam/Proton, Lutris, native); pass it if detection fails

Then manage everything from the Intel tab: tick the channels to monitor (the shipper reports the ones it sees, polls the allowlist ~60s, and ships only those).

anjin-intel status      # installed? running? server reachable? last ship?
anjin-intel uninstall   # stop + remove the service, binary and config

Windows

Install — saves the config, copies the binary to %LOCALAPPDATA%\Programs\anjin, and registers a Task Scheduler logon task so the shipper starts when you log in:

.\anjin-intel.exe install --server https://anjin.example.net --token <enrollment-token>

Expect one UAC prompt. Registering a scheduled task needs elevation even on an Administrator account — a normal terminal holds the UAC-filtered token, where Administrators is deny-only, so schtasks /create answers "Access is denied". Only that one step is elevated; the config and binary are written beforehand as you, so your enrollment token never appears on an elevated command line. Decline the prompt and the install still completes — you just start it yourself and it says so.

The task runs unelevated as you (it has to, to read your Chatlogs), lifts the default 72-hour execution limit that would otherwise kill it every three days, and keeps running on battery.

Run it by hand any time — no flags, it reads the config:

& "$env:LOCALAPPDATA\Programs\anjin\anjin-intel.exe" run

You can skip install and pass the flags each time, but then the token goes on the command line and into PowerShell's history file every launch:

.\anjin-intel.exe run --server https://anjin.example.net --token <enrollment-token>

--logdir is optional: the Chatlogs directory is found by asking Windows where your Documents folder actually is. That matters because the folder is both localized (it's Dokument on a Swedish install) and relocatable — OneDrive moves it under %USERPROFILE%\OneDrive by default on Windows 11 — so there's no single path to hardcode. If you have both a live and an old copy, the one EVE most recently wrote to wins. Pass --logdir to override.

Then pick your channels in the Intel tab, same as Linux. status reports whether the task is registered and whether a shipper is live; uninstall removes the task (another UAC prompt), the binary and the config. Windows won't delete a running .exe, so uninstall says where the leftover is rather than pretending it succeeded.

Run in the foreground (any OS)

anjin-intel run --server <url> --token <tok> [--logdir <EVE/logs/Chatlogs>]

run reads the install config when flags are omitted (that's how the service starts it), and auto-detects --logdir when neither supplies one. It only ships lines written after it starts (no backfill); --channels is just an optional offline/first-run seed. See SPEC.md for the server contract.