Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Agent Quality Gate Pack

    1

    This pack gives your agent three small checks it can run itself, so the claim comes with an exit code and line numbers instead of an opinion.

    $24

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Agent Quality Gate Pack

    Agent Quality Gate Pack

    Example session with this skill installed

    Audit this web PR before merge: check bad.html for layout/CSS defects, scan the Node/Python project for leaked secrets or injection risks, and verify BAD_README.md claims against local files.

    • Read your context and instructions
    • Compiled the agent quality gate

    Real output from running the packaged scripts on small fixtures.

    1. node dom_floor_check.mjs bad.html

    $ node dom_floor_check.mjs bad.html
    dom_floor_check: bad.html
      [warn] line 5 external-dependency: <link> loads a remote resource: //fonts.googleapis.com/css2?family=Inter
          fix: Inline or vendor the asset so the file works offline.
      [error] line 7 html-only-background: Background is set on html but never on body.
          fix: Set the background on body as well (e.g. "html, body { background: ... }").
      [error] line 21 heading-skip: Heading jumps from h1 to h3.
          fix: Do not skip levels; use h2 here.
      [error] line 22 multiple-h1: More than one h1 on the page.
          fix: Keep exactly one h1; demote the others to h2.
      [warn] line 23 inline-font-size: Inline style sets font-size.
          fix: Move the size into a class built on the shared type scale.
      [error] line 24 table-overflow-wrap: Tables present but no "overflow-wrap: anywhere" rule found.
          fix: Add "td, th { overflow-wrap: anywhere; }" so long tokens cannot widen the page.
      [error] line 24 table-mobile-media: Tables present but no "@media (max-width: 560px)" rule found.
          fix: Add a 560px breakpoint that stacks or scrolls the table.
      [error] line 25 img-alt-missing: <img> has no alt attribute.
          fix: Add alt text, or alt="" if the image is purely decorative.
      [warn] line 26 external-dependency: <script> loads a remote resource: //cdn.jsdelivr.net/npm/some-lib
          fix: Inline or vendor the asset so the file works offline.
      [warn] file focus-visible-missing: Interactive elements exist but no :focus-visible style is defined.
    ... (4 more lines)
    (exit 1)
    

    2. node sec_audit_standalone.mjs . (small vulnerable project; secret values are redacted)

    $ node sec_audit_standalone.mjs .
    sec_audit_standalone: 3 file(s) scanned
    [high] .env:1 env-not-gitignored - Environment file is not covered by any .gitignore entry.
        > .env
        fix: Add ".env" and ".env.*" to .gitignore, and rotate any key that was ever committed.
    [high] src/app.js:2 secret-aws-key - AWS access key id.
        > const awsKey = "AKIA****";
        fix: Remove it from source, load it from the runtime environment or a secret store, and rotate it.
    [high] src/app.js:3 secret-jwt - Hard-coded JWT.
        > const sessionJwt = "eyJh****";
        fix: Remove it from source, load it from the runtime environment or a secret store, and rotate it.
    [high] src/app.js:5 cors-wildcard-credentials - Wildcard CORS origin combined with credentials.
        > app.use(cors({ origin: '*', credentials: true }));
        fix: Echo back an allow-listed origin; never combine "*" with credentials.
    [high] src/app.js:7 sql-string-building - SQL built by string interpolation or concatenation.
        > db.query(`SELECT * FROM users WHERE id = ${req.query.id}`);
        fix: Use parameterized queries / bound placeholders.
    [high] src/app.js:8 command-injection - Command string is built dynamically from user input.
        > exec('ls ' + req.query.dir);
        fix: Pass a fixed command with an argument array; validate inputs against an allow-list.
    [high] src/app.js:9 dynamic-code-run - Code is built and run from a string containing user input.
    ... (27 more lines)
    (exit 1)
    

    3. node claim_verifier.mjs BAD_README.md --root .

    $ node claim_verifier.mjs BAD_README.md --root .
    claim_verifier: BAD_README.md
      [warn] line 3 slop-phrase: Stock phrase (seamless): "seamless"
          fix: Replace with the concrete fact it stands in for, or delete it.
      [warn] line 3 slop-phrase: Stock phrase (cutting-edge): "cutting-edge"
          fix: Replace with the concrete fact it stands in for, or delete it.
      [error] line 4 missing-path: Referenced path does not exist: src/ghost.js
          fix: Correct the path, remove the reference, or mark it [U] if the file is meant to appear later.
      [error] line 4 missing-path: Referenced path does not exist: docs/missing.md
          fix: Correct the path, remove the reference, or mark it [U] if the file is meant to appear later.
      [error] line 6 unsupported-claim: Claim without evidence: "All tests passed"
          fix: State the command, exit code or measured value that proves it, or label it [U].
      [error] line 8 unsupported-claim: Claim without evidence: "is complete"
          fix: State the command, exit code or measured value that proves it, or label it [U].
      [warn] line 9 missing-u-label: Hedged statement without [U]: "It should probably work on Windows too."
          fix: Append [U] (unverified) or verify it and cite the evidence.
      [error] line 10 line-out-of-range: Line 999 is past the end of src/real.js
          fix: Point to a line that exists.
      [error] line 12 unsupported-claim: Claim without evidence: "全テストが通過"
          fix: State the command, exit code or measured value that proves it, or label it [U].
    summary: 6 error(s), 3 warning(s)
    (exit 1)
    
    3. node claim_verifier.mjs BAD_README.md

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Stop taking the agent's word for it

    AI coding agents say "done" and "all tests passed" with the same confidence whether or not it is true. This pack gives your agent three small checks it can run itself, so the claim comes with an exit code and line numbers instead of an opinion.

    Each skill ships one self-contained Node.js script (Node 18+, standard library only). It prints findings or results with a fix, and exits non-zero when the gate fails. Add --json for machine-readable output.

    The 3 skills

    1. ui-layout-floor-gate checks one HTML or CSS file:

    • inline font-size styles and too many font sizes
    • background set on html but not body
    • tables without overflow-wrap or a narrow-screen rule
    • skipped heading levels, several h1, images without alt
    • remote CDN, font or script dependencies

    2. offline-ai-app-sec-scan scans JS, TS and Python projects:

    • hard-coded secrets (redacted in the output) and .env files not in .gitignore
    • SQL built by string concatenation, command injection, dynamic code
    • raw HTML sinks, open redirects, wildcard CORS with credentials

    3. anti-slop-spec-verifier checks a PR description, README or report:

    • file paths and line numbers that do not exist
    • claims such as "all tests pass" with no evidence next to them
    • stock phrases and guesses that lack a [U] label

    Built for agents and for review

    • Single file per skill, standard library only: copy one script anywhere and run it
    • Works when the skill is loaded through the Agensi MCP: write the script to a temporary file, run it, delete it
    • No network access, no shell commands, no environment reads
    • Exit code and counts you can quote as evidence

    What is in the ZIP

    • README.md
    • shared/verification-principles.md
    • skills/anti-slop-spec-verifier/references/anti-slop-rubric.md
    • skills/anti-slop-spec-verifier/scripts/claim_verifier.mjs
    • skills/anti-slop-spec-verifier/SKILL.md
    • skills/offline-ai-app-sec-scan/scripts/sec_audit_standalone.mjs
    • skills/offline-ai-app-sec-scan/SKILL.md
    • skills/ui-layout-floor-gate/scripts/dom_floor_check.mjs
    • skills/ui-layout-floor-gate/SKILL.md

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.

    These are static, line-based checks. They do not replace a code review, a penetration test, a browser render, or your own test suite.
    A clean result means "no known pattern found", not "secure" or "correct".
    The verifier checks that a claim has evidence next to it, not that the evidence is true.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed today

    Software
    Node.js 18+

    Offline read-only checks: reads local files only, makes zero network calls, reads no env vars, and never edits your repository.

    Permissions

    Terminal / Shell
    Read Files
    Listedtoday

    What's inside

    Frequently Asked Questions

    No. The scripts only read local files and print to the terminal. They contain no network, shell, or environment-variable code.

    Only Node.js 18 or newer. There is no package install.

    Yes. Each script is a single file with no imports of its siblings.

    Yes. Each SKILL.md explains the temporary-file procedure: write the script out, run it, delete it.

    New in Code Quality & Review