oauth-evidencelisted

Writes OAuth-only headless demo runbook and bench parity evidence templates for the playground—debug preflight, golden runner commands, artifact checklist, known cache-TTL blocker. Use when Alan asks how to demonstrate Pro/Max OAuth via bridge without API keys; do not run live calls unless Alan explicitly asks.
alankatanoisi/claude-local-bridge-playground · ★ 1 · Code & Development · score 57

Install: claude install-skill alankatanoisi/claude-local-bridge-playground

# OAuth Evidence Collector Documents **how to collect policy-grade evidence** that the runner uses Claude Code OAuth through the bridge—**without** API keys and **without** running live calls unless Alan asks. ## When to use - Create or update `lab-notes/parity/oauth-headless-demo-runbook.md` - Create or update `lab-notes/parity/bench-parity-evidence.md` (stub vs `--live` tables) - Preflight checklist: extension, bridge port, debug token, capture proxy confusion - Document known blockers (e.g. cache_control TTL ordering) ## When not to use - Official policy citation refresh (**anthropic-official**) - Matrix adopt/skip without run steps (**parity-archivist**) - Implementing bridge fixes (**default agent**; bridge files gated in CHARTER) ## Charter Read [`lab-notes/agents/CHARTER.md`](../../lab-notes/agents/CHARTER.md). **Default: do not run** `localhost:11437` or live runner unless Alan explicitly asks. ## File ownership | File | Scope | | ---- | ----- | | `lab-notes/parity/oauth-headless-demo-runbook.md` | Demo procedure | | `lab-notes/parity/bench-parity-evidence.md` | Bench commands + metric tables | ## Read first 1. `lab-notes/OAUTH_ONLY_DIRECTION.md` 2. `README.md`, `QUICKSTART.md`, `HEADLESS_AGENT_RUNNER_BEGINNER_GUIDE.md` 3. `docs/threat-model.md` (safety flags for golden command) 4. `src/runner/model-client.js` (default `http://127.0.0.1:11437/v1/messages`) 5. `src/runner/run.js` (`applyCacheControlBudget`) + `src/credentials.js` (`prependClaudeCodeSystem`)