MCP server that provides web automation tools via Puppeteer, including URL fetching and web search capabilities with embedded Chrome DevTools MCP support
Two tools with clear, action-oriented names (fetch_url, search) and reasonable descriptions (120-180 chars). Both have minimal input schemas with only one parameter each. Schemas lack type constraints, enums, and output documentation. No error handling guidance, no parameter validation rules, and no mention of pagination or result limits. Descriptions are adequate but lack dependency hints and prerequisites. Missing security considerations (e.g., untrusted external data handling despite mentioning 'content fence'). No tool annotations (readOnlyHint present in metadata but not in schema).
Fetches a URL using Puppeteer, navigates to the page, waits for network settlement, extracts content via Readability, converts to Markdown, and wraps in a content fence for untrusted external data
Performs a web search using the configured search backend (Google or DuckDuckGo), navigates to the search engine, enters the query, extracts results, and returns them as Markdown
Input schemas lack type constraints and validation rules. Both tools accept only a bare string parameter with no format, length, or pattern constraints. No guidance on valid URL formats, search query length limits, or special character handling.
Output schemas are not documented. LLMs cannot plan downstream operations or extract structured data. fetch_url returns Markdown-wrapped content; search returns Markdown results, but no schema is visible in tool definitions.
No error handling guidance. Descriptions do not explain what happens on network failure, invalid URL, timeout, or search engine unavailability. LLMs have no recovery path.
fetch_url description mentions 'content fence for untrusted external data' but provides no security context. No guidance on sanitization, injection risks, or when to use this tool vs. alternatives.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
No pagination or result limits documented. search tool may return hundreds of results; fetch_url may extract very large pages. No mention of limits, truncation, or how to handle large responses.