Borrowing it
Nothing to install: this file belongs to acartag7/mcp-sso. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/acartag7/mcp-sso/main/.claude/skills/class-closure-review/SKILL.mdgit clone --depth 1 https://github.com/acartag7/mcp-ssoWrote 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/acartag7/mcp-sso/class-closure-review)<a href="https://agentmods.dev/skills/acartag7/mcp-sso/class-closure-review"><img src="https://agentmods.dev/badge/skills/acartag7/mcp-sso/class-closure-review/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/acartag7/mcp-sso/class-closure-review"><img src="https://agentmods.dev/badge/skills/acartag7/mcp-sso/class-closure-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00084 | $0.01148 |
| Opus 5 | $0.00042 | $0.00574 |
| Sonnet 5 | $0.00017 | $0.00230 |
| Haiku 4.5 | $0.00008 | $0.00115 |
Grade A, and why
class-closure-review 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 11d 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 — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Class-closure review (mcp-sso)
Exact-head local review. The unit of work is a behavior, not a
file:line. PASS requires a filled matrix. An empty applicable cell
is FAIL, even if the named instance looks fixed.
A prior CLEAN is not evidence. "Swept siblings" with no cells is FAIL. This skill is how AGENTS.md items 1–4 are executed, not restated.
Do not edit the reviewed tree. Do not implement the fix in this pass.
Mutations belong under /tmp.
Runner
Keep the matrices. Starve the ceremony. Full rules: runner.md.
- After one environmental full-suite failure, classify the
common shape (
EPERMon bind, sandbox, …) and stop rerunning it. - Command output is summaries first. Full logs only for an unexpected failure.
- No web, personal memory, or live-document lookup unless the diff leaves a factual question unresolved.
- Budget 40k–60k tokens. Exceed it only after a genuine
open class. Classify irrelevant axes
n/aimmediately.
Emit the output contract once. Hosted Codex remains the merge gate.
When
- Local review of a PR or exact commit before Codex.
- After a hosted finding, before the next push or re-request.
- Follow-up PRs that are the next sibling of the last merge
(Windows warning trilogies, sanitization then the
{ ok: false }leftover, redirect replace then prepare-vs-approve).
Rules
- Review the exact named commit. Print the full SHA. Any later push makes the result stale.
- Name the defective behavior in one sentence (what still works that the change says cannot). Then fill cells.
- Load matrices.md. Mark every cell
hit/n/a/clean.n/a= that axis is not a sibling of this behavior (one-line reason). Empty = it is a sibling in this tree and you skipped it. Do not invent a P1/P2 for a file the tree does not contain. Do not walk stores or schema because they exist in the checkout. "Every store call is wrapped" means Port methods on an already-open store (find/remove/create/ CAS / getters /destroy).openSqliteStore/ MySQL open / a store constructor is M4 (guard before the factory), not an empty M1 wrap cell, unless the claim literally says store-open errors are sanitized. - Guarantee verbs need enforcing code and a test that goes red if that code is removed. Softening the sentence is not the fix. Grep is in AGENTS.md §7.1 — run it.
- After any finding, do not push until the class is closed. Fixing the named instance and re-requesting Codex is the round multiplier this skill exists to stop.
- If review discovers an edge the contract never named, STOP. Amend the contract or re-cut. See output-contract.md.
What ships with it
22 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.
- evals/class-closed/graders/class-closed.md 710 B
- evals/class-closed/prompt.md 2.1 KB
- evals/CORPUS.md 2.3 KB
- evals/guard-after-open/graders/guard-after-open.md 557 B
- evals/guard-after-open/prompt.md 898 B
- evals/leftover-claim/graders/leftover-claim.md 610 B
- evals/leftover-claim/prompt.md 786 B
- evals/name-not-shape/graders/name-not-shape.md 569 B
- evals/name-not-shape/prompt.md 906 B
- evals/one-call-site/graders/one-call-site.md 553 B
- evals/one-call-site/prompt.md 1019 B
- evals/RESULTS.md 12 KB
- evals/run-isolated.sh 2.5 KB runs code
- evals/starter-not-library/graders/starter-not-library.md 557 B
- evals/starter-not-library/prompt.md 1.0 KB
- evals/stored-not-rechecked/graders/stored-not-rechecked.md 595 B
- evals/stored-not-rechecked/prompt.md 830 B
- README.md 842 B
- references/freeze-a-case.md 898 B
- references/matrices.md 5.7 KB
- references/output-contract.md 2.1 KB
- references/runner.md 2.3 KB
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.
- 11d ago First seen · 105 lines · 84 tokens per session scan A f4f25e379b5b
class-closure-review is a skill published in the GitHub repository acartag7/mcp-sso (8 stars, last pushed today), licensed MIT. It adds 84 tokens to every session and 1,148 once invoked, about $0.0004 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
auth0-docs
Auth0 — identity platform: auth flows, Universal Login, SSO, identity providers, MFA, RBAC, Actions, tokens.
install
Install and configure @brainwebuk/payload-plugin-mcp-oauth — OAuth 2.1 + PKCE + Dynamic Client Registration for a Payload @payloadcms/plugin-mcp server, so it can be added as a Custom Connector in Claude.ai. Use when adding OAuth to a Payload MCP app; wiring the plugin, proxy/middleware, or env; or debugging the…
pr-review
Review a GitHub pull request using multiple expert personas. Takes a PR URL as input, analyzes the changes, and generates comprehensive review feedback from different perspectives (Merge Specialist, Frontend, Backend, Security, DevOps, AI/Agent, SRE, Chief Architect).
release-pr-review
Review pass on an open release PR (release/ → main) — the step between git-wrapup and release-and-publish when a project releases in gated release PR mode. Reads the PR's commit range through the code-simplifier lens plus a correctness review, verifies whatever an automated reviewer left on the PR, lands fixes as…
pixiv-cli-pr
Prepare, create, update, and monitor pixiv-cli pull requests using the repository template, verification checklist, review handoff, and authorization boundaries. Use when drafting a PR body, opening or editing a PR, requesting review, checking readiness, or handing a merged PR to release preparation.
code-review
Review ServiceNow server-side scripts for ES5 violations, ACL/injection/XSS issues, N+1 queries, missing setLimit/error handling, hard-coded sysids, and business-rule recursion risks.