Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Ghost Bridge Debug Playbooks

    1

    When a bug appears on a real page, "open DevTools and look around" wastes your agent's turns.

    Free

    Security scanned
    Ghost Bridge Debug Playbooks

    Ghost Bridge Debug Playbooks

    Example session with this skill installed

    Our staging page comes up blank after login. The console shows nothing obvious. Figure out what's wrong.

    • Read your context and instructions
    • Compiled the ghost bridge debug

    Summary

    Diagnosed the blank page: the entry JS chunk failed to load — the gateway returns a 502 HTML error page instead of JavaScript, so the app never boots and #root stays empty.

    What I did

    1. get_last_error — no JS exception on the page.
    2. list_network_requests — the entry chunk request failed.
    3. get_network_detail on that chunk — status 502, response body is an HTML error page, not JavaScript.

    Fix

    Redeploy the frontend release or fix the gateway route; verify the chunk URL returns application/javascript.

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

    About this skill

    When a bug appears on a real page, "open DevTools and look around" wastes your agent's turns. This skill hands it the exact tool-by-tool playbook for each symptom: what to call first, what the response fields mean, when to stop, and how to finish with a diagnosis report your team can act on.

    Built for and requires ghost-bridge — the free, open-source zero-restart Chrome bridge for MCP (npm install -g ghost-bridge, MIT). Ghost Bridge attaches AI to the Chrome session you are already logged into: no separate browser, no lost login state, no reproduction steps. The skill guides installation on first use.

    What's inside — 11 references and playbooks

    • Triage map: classify any browser symptom in 3 calls
    • Blank/white screen: boot exception → failed entry chunk → hidden app root
    • Dead clicks: four-cause diagnosis (overlay / thrown handler / failed request / silent no-op) with a hit-testing interrogation script
    • Broken layout: screenshot → targeted DOM read → computed-style diff
    • Failing or slow APIs: read real authenticated traffic, replay in page context
    • Production-only bugs: search the deployed bundle, verify hypotheses in the live runtime
    • Web Vitals: metric-driven investigation that names the contributing request or script
    • Multi-tab test loops: bind cases/app tabs, batched checklist runners
    • Round-trip economy: 3-call diagnoses instead of 30-call transcripts
    • Setup & troubleshooting: first-run install, No Bridge recovery, config

    Every playbook ends in a structured report: symptom, diagnosis, evidence, fix, verification.

    Works with Claude Code, Cursor, Codex, or any MCP client that supports skills. Free forever — the playbooks promote ghost-bridge, the open-source tool they are written for.

    Requires ghost-bridge npm package and Chrome extension setup. User must manually click 'Connect' in the extension popup to authorize the session.

    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

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 13 days ago

    • Free with an Agensi account

    Requires

    GitHub
    ghost-bridge
    Chrome

    Permissions

    Terminal / Shell
    Network Access
    Read Files
    Browser

    Allowed Hosts

    github.com
    Listed13 days ago

    What's inside

    Frequently Asked Questions

    Unlike standard tools that launch clean browser instances, this skill attaches to your active, logged-in Chrome session, allowing you to debug complex states and production bugs without losing your session data or credentials.

    This skill requires the ghost-bridge MCP server and the corresponding Chrome extension; it is compatible with MCP-enabled environments such as Claude Code, Cursor, and other IDEs that support local shell and browser automation.

    You will receive a suite of symptom-driven playbooks and optimized execution paths designed to work with the ghost-bridge protocol to diagnose live browser issues efficiently.

    No, this skill is designed for non-destructive debugging, utilizing 'error-first' diagnostics and batched actions to inspect your live site without forcing a page reload or state reset.

    You must have the ghost-bridge npm package installed locally and the Chrome extension active; the skill then uses the `get_server_info` command to verify the connection before beginning a triage session.

    New in LLM Ops