메뉴
Run a focused Centaur tool health smoke test across every live tool CLI. Use when asked to smoke test tools, check all tool auth/connectivity, validate brokered credential injection, or produce a Slack-ready health report for the current deployment.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use Centaur's indexed company context together with direct Slack, Linear, Google Docs, Drive, Calendar, or Granola searches when answering internal company-history, prior-decision, project-context, meeting-context, roadmap/status, or cross-source memory questions. Indexed context includes Slack channels, user-visible Slack DMs, Google Docs, Google Calendar, Linear, and user-visible Granola notes. Use for questions like what was discussed, decided, planned, mentioned, or documented internally, especially when the user did not name one exact source.
Smoke test the full running Centaur system from a new Slack thread. Use when asked to QA the stack, run a smoke test, verify a deployment, check stack health, check deploy readiness, or prove Slack tools, file upload/download, company context, logs, metrics, and tool loading work end to end.
Add, modify, or debug Centaur harness-server backends in crates/harness-server. Use when adding support for a new harness CLI, changing Codex App Server V2 normalization, investigating Claude Code/Amp/Codex streaming or steering behavior, removing Python/TypeScript harness normalizers, or differentially testing real harness stdout/stderr against the shared App Server protocol.
Bring up the full Centaur development stack locally with cargo-run Rust API, Kind agent sandboxes, onepassword/iron-proxy credentials, Tailscale Funnel Slack webhooks, Slack app manifest setup, and centaur-session-cli smoke tests or thread attachment.
Investigate reported auth, credential, permission, API proxy, 401, 403, 503, OAuth, token, or secret-resolution failures by querying VictoriaLogs first. Use when a user says an integration auth failed, a tool got unauthorized/forbidden/service unavailable, an api-proxy request failed, or asks why credentials/secrets are not working.
Scaffold and build new tool integrations in tools/. Use when asked to create a new tool, add an API integration, or build a new client for an external service.
| name | tool-health-smoke |
| description | Run a focused Centaur tool health smoke test across every live tool CLI. Use when asked to smoke test tools, check all tool auth/connectivity, validate brokered credential injection, or produce a Slack-ready health report for the current deployment. |
Use this skill for broad tool-by-tool smoke checks. It is narrower than the full qa skill: it only verifies that each live tool CLI is installed and that its health command succeeds through the normal sandbox credential path.
Do not invent one-off probes for each tool. The canonical smoke surface is:
<tool> health
Run the bundled runner from a sandbox or environment where centaur-tools and tool shims are installed:
uv run .agents/skills/tool-health-smoke/scripts/run_tool_health_smoke.py
The runner discovers tools with centaur-tools json, then executes each <tool> health command with a bounded timeout.
Review the generated Slack-friendly report. The first line is the overall result:
Overall: PASS|FAIL|PARTIAL - <reason>
Return the report in the Slack thread as the assistant response. Do not call the Slack posting API unless the user explicitly asks you to post through the Slack tool.
returncode != 0, invalid JSON, missing ok, or ok: false as a failed tool health check.health command as a failure. Tools are expected to expose health.uv run .agents/skills/tool-health-smoke/scripts/run_tool_health_smoke.py --timeout 45
uv run .agents/skills/tool-health-smoke/scripts/run_tool_health_smoke.py --concurrency 4
uv run .agents/skills/tool-health-smoke/scripts/run_tool_health_smoke.py --only slack,websearch,vlogs
uv run .agents/skills/tool-health-smoke/scripts/run_tool_health_smoke.py --json
Use --json only when you need machine-readable results for follow-up analysis. Use the default text output for Slack.
PASS: every discovered tool health command returned valid JSON with ok: true.FAIL: one or more tools failed, timed out, returned malformed output, or omitted the required health result.PARTIAL: discovery succeeded but no tools were found, or the run was intentionally filtered.When a tool fails because a credential is missing, check whether the client incorrectly requires an environment variable instead of using secret() or proxy-injected auth.