Skip to content

AGENTS.md

Latest commit

 

History

History
48 lines (35 loc) · 3.11 KB

AGENTS.md

File metadata and controls

48 lines (35 loc) · 3.11 KB
Raw

AGENTS.md

Command Code (commandcode.ai) model provider plugin for OpenClaw, with three-tier model resolution: a generated static baseline for pre-credential discovery, a live catalog refreshed at runtime, and a dynamic resolver (resolveDynamicModel) for model ids missing from the per-agent registry.

Setup commands

  • Install deps: npm install
  • Start dev: npm run build (no dev server — plugin is a runtime entry built to dist/)
  • Build: npm run build (tsc -p tsconfig.json)
  • Test: npm test (vitest run)
  • Typecheck: npm run typecheck (tsc --noEmit)

Project layout

  • index.ts — plugin entry: registers the commandcode provider, live/static catalogs, dynamic resolution, transport mapping
  • src/ — provider logic; baseline.models.ts is generated, never hand-edited
  • test/ — Vitest unit tests (transport + projection + dynamic resolution)
  • scripts/ — Node .mjs utilities (generate-baseline.mjs, smoke.discovery.mjs, live-test.mjs)
  • .github/workflows/ — CI (typecheck + build + tests)
  • dist/ — build output, committed: OpenClaw git/package installs require the compiled runtime (./dist/index.js); the TypeScript-source fallback only applies to local dev paths (plugins.load.paths, --link). Regenerate with npm run build before committing.
  • openclaw.plugin.json — manifest (provider id, auth env var, onboarding choice, modelCatalog)
  • package.json — package + openclaw extension metadata (prepack builds dist for npm publish)

Code style

  • TypeScript strict mode (tsconfig.json: strict: true), module/moduleResolution NodeNext, target ES2022
  • ESM ("type": "module"); scripts are .mjs, no .d.ts is shipped intentionally
  • No linter/formatter configured — match surrounding code style manually

Testing instructions

  • Unit tests: npm test (Vitest, test/modelMapping.test.ts) — covers the projection, the dynamic model resolver, and a drift guard that compares the manifest modelCatalog against the baseline projection
  • Discovery smoke: node scripts/smoke.discovery.mjs (hits the live public endpoint)
  • Live inference: node scripts/live-test.mjs (reads the key from ~/.openclaw/secrets/providers.json, never prints it)
  • Baseline refresh: node scripts/generate-baseline.mjs (rewrites src/baseline.models.ts and the modelCatalog block of openclaw.plugin.json from the same projection as projectModel in index.ts — keep the two in sync), then npm run build to refresh dist/
  • CI also fails when the committed dist/ does not match a fresh build — always rebuild before committing source changes
  • Add tests for every new behavior; all tests must pass before opening a PR

PR & commit conventions

  • Branch from main; never push to it directly
  • Commit message: conventional commits (feat: / fix: / docs: / refactor:)
  • Open PR once CI (typecheck + build + test) is green

Security

  • Never commit secrets — .env and .env.* are in .gitignore
  • The API key is read from the COMMAND_CODE_API_KEY env var (or the user's auth profile / provider config) only; never hardcode it