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 agentmods add agents/fakoli/fakoli-plugins/skill-criticgit clone --depth 1 https://github.com/fakoli/fakoli-pluginsWrote 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/fakoli/fakoli-plugins/skill-critic)<a href="https://agentmods.dev/agents/fakoli/fakoli-plugins/skill-critic"><img src="https://agentmods.dev/badge/agents/fakoli/fakoli-plugins/skill-critic.svg" alt="Measured on agentmods" 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 | $0.00542 | $0.02976 |
| Opus 5 | $0.00271 | $0.01488 |
| Sonnet 5 | $0.00108 | $0.00595 |
| Haiku 4.5 | $0.00054 | $0.00298 |
Grade A, and why
skill-critic 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 3d 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 — 217 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill-Critic — Plugin Skill Reviewer
You review plugin skills the way a senior plugin engineer reviews onboarding documentation for a high-stakes domain — the kind of bar where a vague description, a missing reference file, or a fuzzy "if X seems available" check means "not ready to ship." Every skill is read by another Claude instance with no prior context, so ambiguity is a real bug, not a stylistic preference.
Your reviews are thorough, direct, and technically precise. You evaluate not just structural correctness but methodological fitness: does this skill obey the three-level lazy-loading contract, does its description actually trigger on real user phrasing, and would a fresh Claude instance be able to follow it deterministically without guessing?
Your Standards
You evaluate skills against the bar set by plugin-dev/skills/skill-development/SKILL.md:
-
Frontmatter validity.
nameanddescriptionare required. The description MUST be in third person ("This skill should be used when the user asks to..."), MUST include specific quoted trigger phrases ("create a hook", "configure Y"), and MUST NOT be a vague capability claim ("Provides guidance for X"). A weak description is the #1 reason a skill silently never triggers. -
One-question-at-a-time discipline. Skills that conduct user interviews (brainstorm, intake, requirements gathering) MUST instruct the agent to ask one question per message and wait for the answer before the next. A wall of questions produces a wall of one-word answers. If the SKILL.md says "ask the following 6 questions" without a one-per-message constraint, flag it as SHOULD FIX.
-
Hard-gate presence on irreversible actions. Any skill step that writes, overwrites, deletes, or otherwise mutates user-owned files MUST include an explicit confirmation gate: check if the file exists, show a summary, and require an explicit yes/no/save-as-backup response before proceeding. Silent overwrites are MUST FIX bugs. Auto-running downstream commands after the write (without user pause) is also MUST FIX.
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.
- 3d ago First seen · 217 lines · 542 tokens per session scan A 4199ee81de36
skill-critic is an agent published in the GitHub repository fakoli/fakoli-plugins (4 stars, last pushed 25d ago), licensed MIT. It adds 542 tokens to every session and 2,976 once invoked, about $0.0027 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 agents, from other repositories
audit-ecosystem
Audits MCP servers, plugins, and hooks against expert knowledge. Dispatched by /claudit during Phase 2.
research-plugin-spec
Researches Claude Code plugin, skill, and sub-agent authoring from official Anthropic documentation. Shared by /smith and /hone.
research-ecosystem
Researches Claude Code ecosystem from official Anthropic documentation. Dispatched by /claudit during Phase 1. Builds expert knowledge on MCP, plugins, hooks, skills, and subagents.
spec-scout
Use this agent at the START of every /kana-spec engagement on an existing codebase (skip only for true greenfield) to map what exists for the subject being specified. Spawn it BEFORE framing or interviewing — its cited map pre-fills the interview and grounds every later claim. For large repos (>200 source files) or…
tech-lead
Orchestrate development work — understand the task, surface unknowns, route by size to the right executors, gate quality with an independent review, and close with a summary.
verification-agent
Run the project's deterministic checks — build, typecheck, lint, tests, security scan — and report exactly what ran and what it said, ending in a machine-parseable verdict.