Loading...
Loading...
Audit a Feast strategy script before it may run with real money — the runtime's static rules first, then a model read against a fixed checklist — and record the verdict, bound to the file's exact bytes. Use when the user has a strategy script for their Feast agent, asks you to write one, or asks whether a script is safe to run.
npx skill4agent add feastmeme/feast-agent feast-auditFLnode "${FEAST_LEAGUE_RUNTIME:-$HOME/.feast-league/runtime}/src/league/bin.mjs"FL audit <file.mjs>| Rule | Refuses |
|---|---|
| A1 | any direct wallet or process access: |
| A2 | imports of any kind: |
| A3 | network: |
| A4 | secrets: key-shaped hex, seed-phrase-shaped words, API keys, key files |
| A5 | dynamic code and reaching for globals: |
| A6 | unbounded loops ( |
| A7 | the wrong shape: other exports, a tick outside 15–3600 s, code at the top level, over 64 KB |
| A8 | any |
| A9 | (a warning) literal sizes over the per-trade limit |
FL audit <file.mjs> --model-verdict pass --notes "M1: … M2: … M3: … M4: … M5: … M6: … M7: … M8: …"
FL audit <file.mjs> --model-verdict refuse --notes "M2: buys on every tick with no condition …"