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 mthines/agent-skills --skill ci-auto-fixgit clone --depth 1 https://github.com/mthines/agent-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/mthines/agent-skills/ci-auto-fix)<a href="https://agentmods.dev/skills/mthines/agent-skills/ci-auto-fix"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/ci-auto-fix/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/mthines/agent-skills/ci-auto-fix"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/ci-auto-fix.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
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 →
- high Tool Misuse · line 230 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- medium Excessive Agency · line 68 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00060 | $0.01824 |
| Opus 5.5 | $0.00024 | $0.00730 |
| Sonnet 5.5 | $0.00012 | $0.00365 |
| Haiku 4.5 | $0.00006 | $0.00182 |
Grade A, and why
ci-auto-fix 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 6d 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 — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CI Auto-Fix
Find the cause, make the smallest justified fix, and verify CI on the resulting revision. Investigation scales with uncertainty and blast radius; completion checks apply to every path. Maximum four fix-push cycles per invocation.
Read anti-patterns first. Load other references only at the decision they support. Reuse evidence from a caller when its revision, run attempt, and scope match; do not repeat discovery to satisfy phase labels.
Phase 0 — Resolve access and target
Resolve GitHub access once using
github-access. Commands below
show the gh path; use that reference's mapping for MCP and report capability gaps.
Accept a PR URL, Actions run URL/ID, or check-run ID. With no argument, resolve the current branch's open PR; if none exists, look for a failed run on its current SHA. Ask only if the target remains ambiguous. Do not choose an old failed run when the current revision is already green. Resolve check-run IDs to their run; do not assume check and workflow IDs are interchangeable.
Record repository, branch, PR (if any), baseline SHA, run IDs/attempts, and local working-tree state. Verify that the checkout matches the target before editing. For a fork, identify the head repository and push destination explicitly. Preserve unrelated local changes; use an isolated worktree when needed. State the target.
Phase 1 — Capture and group failures
List checks and jobs before fetching logs. Capture the complete baseline: failed, passed, pending, skipped, and blocked jobs, including workflow and matrix identity. For each distinct failing job, fetch logs once per run attempt, save them locally, and extract the failing command, error, and enough surrounding context to explain it. Expand excerpts if the cause remains unclear; do not dump entire logs into context.
Group matching failures by likely root cause. Inspect one representative of a repeated signature, then confirm the other jobs share its command/environment; a different runtime or signature deserves its own investigation. Record which failures each proposed fix covers. Independent justified fixes may share a push; do not serialize matrix duplicates into separate CI cycles.
What ships with it
7 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.
- 6d ago Changed · -3 lines · -39 tokens per session f4ff2cf2cea1
- 1mo ago Changed · -198 lines · -55 tokens per session d9f5a9df3142
- 1mo ago Changed · +1 lines 5891cbcdf03e
- 1mo ago First seen · 369 lines · 154 tokens per session scan A 1b8072b234a8
ci-auto-fix is a skill published in the GitHub repository mthines/agent-skills (20 stars, last pushed yesterday), licensed MIT. It adds 60 tokens to every session and 1,824 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
ego-lite-simplify
Find and implement evidence-backed simplifications in the ego-lite repository. Use when reviewing the codebase for dead code, duplicated state or APIs, speculative abstractions, unnecessary compatibility layers, hand-written infrastructure, excessive tests or documentation, or when the user asks to reduce code size or…
gh-fix-ci
Use when a user asks to debug or fix failing GitHub PR checks that run in GitHub Actions; use gh to inspect checks and logs, summarize failure context, draft a fix plan, and implement only after explicit approval. Treat external providers (for example Buildkite) as out of scope and report only the details URL.
comet-github-ci-triage
A workflow for diagnosing failed GitHub Actions checks on a Comet pull request. GitHub Actions runs automated builds and tests; a pull request is a proposed code change waiting to be reviewed and merged.
k8s-yaml-validator
Comprehensive toolkit for validating, linting, testing Kubernetes YAML resources. Use this skill when validating Kubernetes manifests, debugging YAML syntax errors, performing dry-run tests on clusters, working with Custom Resource Definitions (CRDs) that require documentation lookup.
ci-cd-and-automation
Guides CI/CD pipeline design — quality gates, GitHub Actions for Go/TypeScript/PHP, feature-flag-decoupled deploys, secrets scoping, and pipeline optimization. Use when setting up or modifying a build/test/deploy pipeline, adding an automated quality gate, configuring CI secrets, or diagnosing a slow or flaky…
debugging-and-error-recovery
Enforces a stop-the-line triage protocol across failure classes — test failures, build breaks, runtime errors, and production incidents — using bisection, safe-fallback design, and instrumentation lifecycle management. Use when any unexpected failure appears and the question is "what class of failure is this and…