Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add baphuongna/pi-crew --skill async-worker-recoverygit clone --depth 1 https://github.com/baphuongna/pi-crewWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/baphuongna/pi-crew/async-worker-recovery)<a href="https://agentmods.dev/skills/baphuongna/pi-crew/async-worker-recovery"><img src="https://agentmods.dev/badge/skills/baphuongna/pi-crew/async-worker-recovery/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/baphuongna/pi-crew/async-worker-recovery"><img src="https://agentmods.dev/badge/skills/baphuongna/pi-crew/async-worker-recovery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium MCP Rug Pull · line 58 npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.Fix: Pin the version: npx @scope/[email protected]
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00035 | $0.00625 |
| Opus 5 | $0.00017 | $0.00313 |
| Sonnet 5 | $0.00007 | $0.00125 |
| Haiku 4.5 | $0.00003 | $0.00063 |
Grade A, and why
async-worker-recovery scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
async-worker-recovery
Use this skill when a pi-crew run is stuck, stale, interrupted, or has dead workers.
Source patterns distilled
- pi-subagents async patterns: detached runner, status files, result watcher, stale PID reconciler
- pi-crew runtime:
src/runtime/background-runner.ts,async-runner.ts,heartbeat-watcher.ts,worker-heartbeat.ts,crash-recovery.ts,stale-reconciler.ts,deadletter.ts,delivery-coordinator.ts - UI recovery controls:
src/ui/run-dashboard.ts,src/ui/dashboard-panes/health-pane.ts,src/ui/run-action-dispatcher.ts
Rules
- Distinguish historical dead-heartbeat events from current active failures. Check manifest/task status and event timestamps.
- Heartbeat warnings should only apply to currently running/waiting work, never terminal runs/tasks.
- Stale reconciliation order: result/terminal evidence → PID liveness → stale threshold/active evidence.
- Reconcile state under run lock and re-read inside the lock before repair.
- Deadletter entries are evidence, not automatic proof of permanent failure; inspect attempts and later completion events.
- For background runs, verify PID liveness and background log before declaring stuck.
- Session delivery should queue while inactive and flush only to the current generation/session.
- Do not poll in sleep loops waiting for async completion if the system has a watcher/result notification path.
Operator checklist
- Load manifest/tasks and recent events.
- Check
manifest.async.pidand process liveness. - Check heartbeat
lastSeenAt, progresslastActivityAt, and terminal status. - Inspect deadletter and diagnostic report.
- Choose recovery: resume, retry, kill stale, diagnostic, or no-op historical notification.
Enforcement — Worker Recovery Gate
Before taking recovery action, verify:
- Run status is not terminal (completed/failed/cancelled)
- Heartbeat is genuinely stale (not just delayed polling)
- PID is dead or stale threshold exceeded
- Recovery action matches run state (resume vs retry vs kill)
- Session generation matches before state modification
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 62 lines · 35 tokens per session scan A cfd4a0ffb495
async-worker-recovery is a skill published in the GitHub repository baphuongna/pi-crew (52 stars, last pushed yesterday), licensed MIT. It adds 35 tokens to every session and 625 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
printing-press-amend
Amend a published CLI from one of two input sources: (1) dogfood mode mines the active Claude Code session transcript for friction (missing flags, hand- rolled API payloads, silent-null returns); (2) direct-input mode accepts user-supplied asks (rename a command, add commands or feeds, fix a named bug, optionally…
convex-performance-audit
Audits Convex performance for reads, subscriptions, write contention, and function limits. Use for slow features, insights findings, OCC conflicts, or read amplification.
convex-insights
Query a running Convex app's logs + health in natural language (official MCP): failures, slow/expensive functions, deploy causality — scoped, evidence-backed, with a dashboard deep link.
ssl-proxy-troubleshoot
Systematic workflow for troubleshooting SSL/proxy connectivity issues with government websites.
diagnose-backend-bug
Diagnose a bounded backend or multi-service failure from GitHub Issues, Jira, Aone, user-provided exports, logs, traces, responses, stack traces, or job records. Use when a service, API, RPC, worker, queue, CLI, or scheduled job bug needs correlation through the project's existing observability route before repair; do…
platform-apex-anonymous-run
Use this skill to run anonymous Apex against the connected Salesforce org — from a .apex file or a pasted snippet — capturing the debug log, surfacing compile and runtime errors, and summarizing results. Trigger on phrases like "run this anonymous apex", "execute this script against my org", "run this snippet of…