Loading ToolBench

Reading the current corpus…

CrawlIt MCP server · F · ToolBench

← Browse

CrawlIt

github.com/FBR65/CrawlIt

Web crawling and conversion system with FastAPI and MCP server support

<=2025-11-25STDIO1 tools

49/100

Definition Quality60

Protocol Readiness47

Supportability33

Spec posture

Inferred effective spec <=2025-11-25·current baseline 2026-07-28

Static source inference · medium confidence · detected: Logging

Deprecated protocol patterns detected

Summary

Single tool 'crawl' has a clear verb-noun name and proper JSON Schema with typed parameters. However, the tool description is minimal (67 chars, below the 194-char baseline), lacks WHEN/WHY guidance, and does not explain the whitelist validation behavior or error recovery paths. Parameter descriptions are present but generic ('URL to crawl' lacks format/constraint details). No output schema is documented, the response is inferred as markdown text but not formally specified. Error handling is basic (ValueError wrapping) with no recovery guidance. The tool is READ_ONLY (safe) but lacks idempotent/destructive annotations. Overall: functional but below production baseline.

Tools (1)

crawlread onlysource verified60/100

Crawl a URL and return filtered content as markdown

Issues (5)

HIGHtool-description ↗

Tool description is too brief (67 chars vs 194-char baseline). Lacks WHEN to use, prerequisites (whitelist requirement), and what the output contains. LLMs cannot determine selection criteria.

crawl

HIGHresponse-shaper ↗

Output schema is not documented. Response is inferred as markdown text but no formal schema (type, fields, length limits) is provided. LLMs cannot plan downstream operations or validate results.

crawl

MEDIUMconstrained-input ↗

Parameter 'url' description lacks format/constraint details. Should specify: valid URL format, whitelist requirement, and example domains. Current description 'URL to crawl (must be in whitelist)' is vague about what domains are allowed.

crawl

MEDIUMrecovery-guide ↗

Error handling returns generic ValueError messages without recovery guidance. 'Failed to crawl URL: ...' tells LLM nothing about retry eligibility, user-fixable issues, or next steps.

crawl

Recommendations

Expand tool description to 150 - 250 chars: 'Crawl a URL and extract its main content as markdown. Supports only whitelisted domains (e.g. example.com, docs.example.com). Returns cleaned markdown with optional navigation elements. Use this to fetch and summarize web pages for the user.'
Document output schema explicitly: 'Returns a TextContent object with type="text" and text field containing markdown (max 50KB). Includes title, main content, and optionally navigation links.'
Add 'readOnlyHint': true to the tool schema to declare it is safe to call without side effects.
Enhance 'url' parameter description: 'A valid HTTP/HTTPS URL. Must be from the whitelisted domain list (e.g. example.com, docs.example.com). Invalid domains return a 400 error with available options.'
Improve error messages with recovery guidance: Instead of 'Failed to crawl URL: Domain not in whitelist', return 'Domain "example.org" not in whitelist. Allowed domains: example.com, docs.example.com. Contact admin to add new domains.' This guides the LLM to ask the user or suggest alternatives.

Spec posture evidence

Inferred effective spec: <=2025-11-25.

Relies on Logging (deprecated) - log to stderr or use OpenTelemetry

Score history

Overall score trend

First recorded score · v2 rubric

49/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	49	<=2025-11-25	v2