Enhanced CDP-native MCP debugging platform with 85+ resources and 40+ tools for React applications. Provides AI-assisted debugging through Model Context Protocol with Chrome DevTools Protocol integration.
11 tools with schemas present but significant quality gaps. Tool names follow verb_noun pattern (apollo_*, cdp_*) but descriptions are minimal (10-30 chars, well below 50-200 char baseline for A+ tools). Parameters lack descriptions in most cases, only 'sessionId' and primary params have brief text. No output schemas documented. Error handling absent. Apollo tools expose GraphQL internals without guidance on when to use each variant. CDP tools lack safety warnings for destructive operations (clear_cookies, reload). No idempotency hints or recovery guidance.
Clear Apollo Client cache
Inspect Apollo Client cache
Refetch Apollo Client query
Write data directly to Apollo cache
Clear all cookies
Evaluate JavaScript expression in the browser
Get all cookies
Descriptions critically short (10-30 chars). Rubric baseline for A+ tools is 50-200 chars. LLMs cannot determine when to select apollo_inspect_cache vs apollo_refetch_query vs apollo_write_cache without richer context.
No output schemas documented. Rubric requires 100% of A+ tools to have documented return types. Agents cannot plan downstream calls or extract required fields without knowing response structure.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | <=2025-11-25 | v2 |
Navigate to a URL
Reload the current page
Take a screenshot of the current page
Set a cookie
Destructive operations (apollo_clear_cache, apollo_write_cache, cdp_clear_cookies, cdp_reload) lack confirmation/dry-run patterns and safety warnings. No error guidance for recovery.
Parameter descriptions missing or trivial. 'sessionId' described as 'Chrome session ID (optional)' but no guidance on how to obtain it, what happens if omitted, or when it's required vs optional.
No error handling or recovery guidance. Tools return success/failure but do not tell LLM what to do next (retry, ask user, abort). Rubric requires 'Error responses must tell the LLM what to do next.'