Claude Skill

chrome-devtools

Debug Chrome with DevTools MCP: console, network, performance, accessibility.

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download fmind-dot-skills_chrome-devtools-3bbe830.zip · 1 KB

Install

skills CLI npx skills add https://github.com/fmind/dot/tree/main/skills/chrome-devtools
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install fmind-dot@llmmart
Git git clone https://github.com/fmind/dot.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole fmind/dot collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Chrome DevTools

Live Chrome diagnostics via DevTools MCP or its CLI (playwright owns repeatable E2E journeys; web-frontend owns UI implementation).

Setup

Requires Node.js LTS, npm/npx, and Chrome. Reuse an active MCP server if connected, default to the CLI below when none is configured (dot deploys none), or register MCP via mcp-setup:

npx --yes chrome-devtools-mcp@1.13.0 --isolated --headless --no-usage-statistics --no-performance-crux --redact-network-headers

Keep debugging on loopback in a dedicated profile (pages, bodies, cookies, screenshots, and traces may hold sensitive data). --no-usage-statistics and --no-performance-crux disable telemetry and CrUX URL lookups; --redact-network-headers masks non-allowlisted headers (Cookie, Set-Cookie, Authorization, ETag, Content-Length) unless deliberately omitted.

CLI without MCP

  1. Run via pinned package: npm exec --yes --package=chrome-devtools-mcp@1.13.0 -- chrome-devtools <command> (check --help and start --help).
  2. Respect running daemons: check status first; never stop or hijack another task's daemon.
  3. Start scoped: start --workspace="$PWD" --no-usage-statistics --no-performance-crux (auto-start on first tool call grants unrestricted file access).
  4. Tear down: close task-owned pages and stop only daemons started by this task.

Workflow

  1. Select target and snapshot: pick the page ID and take a fresh snapshot before using element UIDs (navigation/rerenders invalidate them).
  2. Console and network: list bounded console logs and failed requests first, fetching only needed request IDs/bodies.
  3. Performance and Lighthouse: pass filePath to performance_start_trace / performance_stop_trace instead of dumping raw traces; lighthouse_audit checks a11y, SEO, and best practices (reloads unless mode="snapshot"; excludes performance).
  4. Memory: take_heapsnapshot is enabled by default; compare_heapsnapshots and deeper analysis require --memoryDebugging on MCP.
  5. Clean up: close task-owned pages after verification.

Gotchas

  • Explicit browser mode: pass headed/headless explicitly (the CLI defaults to headless + --memoryDebugging, which may differ from an MCP config).
  • Workspace write roots: MCP file tools write only inside temp dirs, client MCP roots, or --workspace=<dir> (is not within any of the configured workspace roots otherwise).

Official Skills

Upstream: ChromeDevTools/chrome-devtools-mcp skills (DISABLE_TELEMETRY=1 skills add ChromeDevTools/chrome-devtools-mcp --list --json). Select specific skills (chrome-devtools-cli, a11y-debugging, memory-leak-debugging) under vendor-skill policy; preview—never install—same-name chrome-devtools (name collisions).

Documentation

Files (dot)
  • SKILL.md 3.9 KB
    ---
    name: chrome-devtools
    description: "Debug Chrome with DevTools MCP or its CLI: console, network, performance, accessibility."
    license: MIT
    metadata:
      kind: connector
      author: Médéric HURIER (Fmind)
      source: github.com/fmind/dot/tree/main/skills/chrome-devtools
      created: "2026-09-03"
      updated: "2026-11-11"
    ---
    
    # Chrome DevTools
    
    Live Chrome diagnostics via DevTools MCP or its CLI ([playwright](../playwright/SKILL.md) owns repeatable E2E journeys; [web-frontend](../web-frontend/SKILL.md) owns UI implementation).
    
    ## Setup
    
    Requires Node.js LTS, `npm`/`npx`, and Chrome. Reuse an active MCP server if connected, default to the CLI below when none is configured (`dot` deploys none), or register MCP via [mcp-setup](../agent-harnesses/references/mcp-setup.md):
    
    ```bash
    npx --yes chrome-devtools-mcp@1.13.0 --isolated --headless --no-usage-statistics --no-performance-crux --redact-network-headers
    ```
    
    Keep debugging on loopback in a dedicated profile (pages, bodies, cookies, screenshots, and traces may hold sensitive data). `--no-usage-statistics` and `--no-performance-crux` disable telemetry and CrUX URL lookups; `--redact-network-headers` masks non-allowlisted headers (`Cookie`, `Set-Cookie`, `Authorization`, `ETag`, `Content-Length`) unless deliberately omitted.
    
    ## CLI without MCP
    
    1. **Run via pinned package**: `npm exec --yes --package=chrome-devtools-mcp@1.13.0 -- chrome-devtools <command>` (check `--help` and `start --help`).
    1. **Respect running daemons**: check `status` first; never stop or hijack another task's daemon.
    1. **Start scoped**: `start --workspace="$PWD" --no-usage-statistics --no-performance-crux` (auto-start on first tool call grants unrestricted file access).
    1. **Tear down**: close task-owned pages and `stop` only daemons started by this task.
    
    ## Workflow
    
    1. **Select target and snapshot**: pick the page ID and take a fresh snapshot before using element UIDs (navigation/rerenders invalidate them).
    1. **Console and network**: list bounded console logs and failed requests first, fetching only needed request IDs/bodies.
    1. **Performance and Lighthouse**: pass `filePath` to `performance_start_trace` / `performance_stop_trace` instead of dumping raw traces; `lighthouse_audit` checks a11y, SEO, and best practices (reloads unless `mode="snapshot"`; excludes performance).
    1. **Memory**: `take_heapsnapshot` is enabled by default; `compare_heapsnapshots` and deeper analysis require `--memoryDebugging` on MCP.
    1. **Clean up**: close task-owned pages after verification.
    
    ## Gotchas
    
    - **Explicit browser mode**: pass headed/headless explicitly (the CLI defaults to headless + `--memoryDebugging`, which may differ from an MCP config).
    - **Workspace write roots**: MCP file tools write only inside temp dirs, client MCP roots, or `--workspace=<dir>` (`is not within any of the configured workspace roots` otherwise).
    
    ## Official Skills
    
    Upstream: [ChromeDevTools/chrome-devtools-mcp skills](https://github.com/ChromeDevTools/chrome-devtools-mcp/tree/main/skills) (`DISABLE_TELEMETRY=1 skills add ChromeDevTools/chrome-devtools-mcp --list --json`). Select specific skills (`chrome-devtools-cli`, `a11y-debugging`, `memory-leak-debugging`) under [vendor-skill policy](../agent-project/references/vendor-skills.md); preview—never install—same-name `chrome-devtools` ([name collisions](../agent-project/references/vendor-skills.md#name-collisions)).
    
    ## Documentation
    
    - [Chrome DevTools for agents](https://github.com/ChromeDevTools/chrome-devtools-mcp) · [Tool reference](https://github.com/ChromeDevTools/chrome-devtools-mcp/blob/main/docs/tool-reference.md) · [CLI](https://github.com/ChromeDevTools/chrome-devtools-mcp/blob/main/docs/cli.md)
    - Releases: [chrome-devtools-mcp](https://github.com/ChromeDevTools/chrome-devtools-mcp/releases)
    - Companion skills: [mcp-setup](../agent-harnesses/references/mcp-setup.md), [playwright](../playwright/SKILL.md), [web-frontend](../web-frontend/SKILL.md), [performance](../python-stack/references/performance.md).
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.