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 cacheforge-ai/cacheforge-skills --skill swarm-self-healgit clone --depth 1 https://github.com/cacheforge-ai/cacheforge-skillsWrote 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/cacheforge-ai/cacheforge-skills/swarm-self-heal)<a href="https://agentmods.dev/skills/cacheforge-ai/cacheforge-skills/swarm-self-heal"><img src="https://agentmods.dev/badge/skills/cacheforge-ai/cacheforge-skills/swarm-self-heal/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/cacheforge-ai/cacheforge-skills/swarm-self-heal"><img src="https://agentmods.dev/badge/skills/cacheforge-ai/cacheforge-skills/swarm-self-heal.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00032 | $0.00585 |
| Opus 5 | $0.00016 | $0.00293 |
| Sonnet 5 | $0.00006 | $0.00117 |
| Haiku 4.5 | $0.00003 | $0.00059 |
Grade A, and why
swarm-self-heal 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 9d 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.
What it actually says
When to use this skill
Use this skill when the user wants to:
- Diagnose why a multi-agent swarm feels "stuck" or partially offline
- Check gateway + channel + lane liveness in one run
- Perform bounded auto-recovery (restart + retry only)
- Capture auditable receipts for incident timelines
- Keep a primary watchdog lane plus a backup lane in place
Commands
# Install/refresh watchdog scripts + cron wiring
bash skills/swarm-self-heal/scripts/setup.sh
# Run an immediate canary check
bash skills/swarm-self-heal/scripts/check.sh
# Run watchdog directly (uses deployed workspace path)
bash ~/.openclaw/workspace-studio/scripts/anvil_watchdog.sh
# Optional: increase lane ping timeout for slower providers
PING_TIMEOUT_SECONDS=180 bash ~/.openclaw/workspace-studio/scripts/anvil_watchdog.sh
What it checks
- Gateway health via
openclaw health - Channel readiness via
openclaw channels status --json --probe - Passive lane recency via
openclaw status --json(latest OpenClaw-compatible) - Active lane probe only when stale for
main,builder-1,builder-2,reviewer,designer - Bounded recovery with a single restart pass + targeted re-probe of infra failures
Output contract
The watchdog output includes:
timestamptargetsok_agentsfailed_agentsactionsVERDICTRECEIPT
Safety model
- Bounded recovery only (single restart pass per run)
- No destructive state wipes
- No blind reinstall behavior
- Recovery actions are explicit in output
Notes
- Cron wiring sets both primary and backup watchdog lanes to
xhighthinking. - Telegram target is auto-derived from config when available, with a safe fallback.
- Healthy runs can be summarized as a single line to reduce operator noise.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 9d ago First seen · 78 lines · 32 tokens per session scan A e35053aeb46d
swarm-self-heal is a skill published in the GitHub repository cacheforge-ai/cacheforge-skills (10 stars, last pushed 6mo ago), licensed MIT. It adds 32 tokens to every session and 585 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 2026-08-31.
Other skills, from other repositories
design-incident-response
Use when creating or improving an incident response process for a production system or operational team.
k8s-incident-triage
Read-only Kubernetes incident triage using kubectl. Accepts natural language or structured input. Produces root-cause hypotheses, evidence, and next-step commands.
feishu-troubleshoot
A Feishu troubleshooting guide and diagnostic command for finding problems with the Feishu plugin. Feishu, also called Lark, is a workplace collaboration platform.
kubernetes
Kubernetes operations playbook for deploying services: core objects, probes, resource sizing, safe rollouts, and fast kubectl debugging.
epistemic-graph-troubleshooting
Diagnose and recover a live epistemic-graph engine (the AI-native database): engine down / socket refused, host-daemon crash-loop, circuit-breaker open, backend-type errors, zombie containers, duplicate Swarm tasks, slow first boot. Use when the KG is unreachable, ingestion stalls, go tools error, or a deploy left the…
datadog
A Datadog log-investigation tool for examining production systems. Datadog is a monitoring service that collects application logs, error reports, alerts, and measurements.