Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Doc Audit Contributor

    1

    Find verified technical documentation errors and generate ready-to-submit GitHub PR packages.

    Free

    1 useSecurity scanned

    Doc Audit Contributor

    Example session with this skill installed

    Audit the OpenAI Agent SDK docs for any broken code examples or missing setup steps. I want to make a quick open-source contribution.

    • Read your context and instructions
    • Compiled the doc audit contributor

    I have audited the OpenAI Agent SDK documentation (root: /docs/agent-sdk). After analyzing the Quickstart, Configuration, and Examples pages, I found one critical code error.

    Documentation Audit Findings

    Quickstart Guide (/docs/agent-sdk/quickstart)

    Type: Broken code
    Problem: The initialization example uses AgentClient(api_key=...) but the library actually requires AgentClient(auth_token=...).
    Why: Internal consistency check; the API reference page for AgentClient confirms auth_token is the required parameter.
    Fix: Update the parameter name in the initialization block to auth_token.

    Branch: docs/fix-agent-client-init-param
    Commit message: docs: fix incorrect initialization parameter in quickstart
    PR title: docs: update AgentClient parameter to auth-token in Quickstart
    PR description:

    
    ## What
    The Quickstart guide uses an incorrect parameter name for the AgentClient constructor.
    
    ## Why
    Running the example as-written results in a TypeError: __init__() got an unexpected keyword argument 'api_key'. The core library expects 'auth_token'.
    
    ## Fix
    Changed `api_key` to `auth_token` in the Python code example.
    
    ## Verification
    Cross-referenced with the API Reference documentation and validated against the SDK constructor signature.
    

    Next steps

    1. Create the branch docs/fix-agent-client-init-param in your local clone.
    2. Apply the one-line fix to docs/quickstart.md.
    3. Push and open the PR using the generated description.

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

    About this skill

    Documentation bugs are the easiest way to start contributing to open source, but manually auditing pages for broken code is slow and tedious. This skill automates the discovery of high-value documentation errors that maintainers actually want to fix. It bypasses superficial UI bugs to focus strictly on content accuracy and code usability.

    What it does

    • Broken code detection identifies syntax errors, wrong method names, or incorrect parameters in documentation examples.
    • Dependency verification flags missing imports, environment variables, or setup steps required to run examples.
    • Content clarity audits pinpoint misleading explanations or logical gaps that confuse new developers.
    • Automated PR packaging generates branch names, conventional commit messages, and structured PR descriptions for every finding.

    How it works

    1. Target Identification: Specify a documentation URL or project name to locate the docs root.
    2. Page Selection: The skill crawls the navigation to pick 5 to 8 high-impact pages like Quickstarts and API references.
    3. Deep Content Analysis: It extracts code blocks and prose, simulating a top-to-bottom execution of every instruction.
    4. Validation: It cross-references findings against the source repo or internal consistency checks to ensure the bug is real.
    5. Ready-to-Submit Output: You receive a formatted report with everything needed to open a GitHub PR immediately.

    Frameworks & tools

    This skill requires an MCP server with Playwright-style browser automation tools. It is optimized for CLI-based agents like Claude Code, Codex, or Gemini CLI that can navigate live documentation sites.

    Why this beats prompting it yourself

    Generic prompts often return "style suggestions" or "typo fixes" that maintainers ignore. This skill uses a strict verification logic to find functional errors in code examples, providing the exact PR metadata needed to contribute in seconds.

    Use cases

    • Find "good first issues" in a new open-source library you want to support.
    • Audit your own project's docs before a major release to ensure examples work.
    • Quickly spot breaking changes in documentation after an API update.
    • Scale open-source contributions by automating the discovery and drafting phase.

    Known limitations

    Does not report UI bugs, broken links, accessibility issues, or styling errors. Requires a browser-enabled MCP environment to function.

    Focuses only on content and code accuracy. Ignores UI, links, and styling. Requires Playwright/browser MCP tools. Default depth is 5-8 pages.

    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 any payment.

    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

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Permissions

    Browser
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    Unlike generic LLM prompts that focus on typos or grammar, this skill specifically simulates code execution to identify functional errors, missing dependencies, and incorrect method names that would cause a developer's environment to fail.

    This skill is designed for CLI-based agents like Claude Code or Gemini CLI and requires an environment with an MCP (Model Context Protocol) server equipped with browser automation tools like Playwright.

    You receive the specialized skill logic for documentation auditing, pre-configured workflows for PR generation, and a guide on connecting it to your browser-enabled MCP server.

    No, the skill focuses strictly on content accuracy and code usability; it does not check for broken hyperlinks, UI/UX bugs, or CSS styling issues.

    The skill is optimized for high-impact technical pages such as Quickstarts, Tutorials, and API References, crawling 5 to 8 pages per run to ensure deep, high-quality analysis.

    The skill generates a structured report including suggested branch names, conventional commit messages, and a detailed description of the fix, allowing you to open a PR in seconds.

    New in Workflow Automation