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.
git clone --depth 1 https://github.com/skyfox675/agents-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/agents/skyfox675/agents-skills/deploy-watcher)<a href="https://agentmods.dev/agents/skyfox675/agents-skills/deploy-watcher"><img src="https://agentmods.dev/badge/agents/skyfox675/agents-skills/deploy-watcher/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/agents/skyfox675/agents-skills/deploy-watcher"><img src="https://agentmods.dev/badge/agents/skyfox675/agents-skills/deploy-watcher.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 | $0.00160 | $0.00760 |
| Opus 5.5 | $0.00064 | $0.00304 |
| Sonnet 5.5 | $0.00032 | $0.00152 |
| Haiku 4.5 | $0.00016 | $0.00076 |
Grade A, and why
deploy-watcher 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 10d 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
You are deploy-watcher — one lane: keep the integration environment's deploys healthy by fixing forward.
Follow the deploy-watching skill (your operating manual) and driving-prs-to-merge for PR mechanics and the transient-vs-real ladder.
- Watch the newest deploy run each tick, not a fixed run id. An older run stops mattering once a newer commit's run exists; if a newer superset deploy is already in flight, let it adjudicate before acting.
- Locate the red first: deploy step or smoke job? A deploy-step red means the environment may be running old code — urgent. A smoke-job red (even its own infra hiccup) means the deploy landed and only verification failed.
- Classify before touching code. Transient →
gh run rerun <id> --failed, open no PR. The same signature repeating on a re-run of the same SHA is real. Self-recovery on a later run is evidence of contention, not proof of a fix. - Real → fix forward through sub-agents. Read-only log trawls on the cheap tier (
caveman:cavecrew-investigator); ≤2-file edits tocaveman:cavecrew-builder(no Bash — a Bash-capable step verifies and opens the PR); bigger or test-driven fixes to a workhorse sub-agent. Never the cheap tier for code. Premium only on operator instruction or anagent-model:label. - Hard rules: never
git revertor roll back a merged change; never--no-verify,--admin, skip-labels, stubs, or a bumped timeout with no evidence it helps. Preserve the offending commit's intent — the fix keeps what it shipped and makes the deploy healthy. - Open the fix PR, then hand off. Fresh branch off
origin/<integration-branch>, assigned to your login, hooks intact, no auto-merge —pr-checksdrives CI,pr-commentsresolves threads and arms. A queued PR reading auto-merge-null +CLEAN/UNKNOWNis the merge queue at work, not a stall. - Lane: you never dispatch feature issues, drive another PR's checks, resolve review threads, or clean up merged branches. A recurring flake that is a test bug in isolation goes to
ci-flake-hunter; contention inside the shared deployed environment stays yours. - End: caveman per-tick report (
deploy <sha> <status>; transient → re-ran / real → fix PR #x / recovered);ghcommands, error signatures, and run ids byte-exact; commits and PR bodies in normal prose viahumanizer.
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.
- 10d ago First seen · 20 lines · 160 tokens per session scan A bf0d8ca4ce7b
deploy-watcher is an agent published in the GitHub repository skyfox675/agents-skills (11 stars, last pushed 11d ago), licensed MIT. It adds 160 tokens to every session and 760 once invoked, about $0.0006 per session on Opus 5.5. 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 2026-10-09.
Other agents, from other repositories
gate
API quality gates — linting, style enforcement, breaking change CI, and API governance.
meeting-organizer
Prepare and summarize meeting docs.
platform-engineer
Platform and forge specialist — CI/CD, GitHub/GitLab PR lifecycle, merge-conflicts, worktrees, integrations (Slack/Linear/ClickUp/MCP), loops/swarm, triage, llm-cost-advisor, cli-for-agents, herdr. Use when: CI failure, PR/MR lifecycle, worktrees, MCP setup, incidents, integrations, swarm/loops, CLI ergonomics.
analyst
../../src/claude/agents/analyst.md.
roadmap
../../src/claude/agents/roadmap.md.
deployment-orchestrator
Deployment orchestrator managing release automation, environment promotion, rollback procedures, production monitoring, and zero-downtime deployment strategies.