Loading ToolBench
Browse FindAgent's public catalog of vetted MCP "doer" agents. Connect the hosted server at mcp.findagent.cloud/mcp for the full authenticated toolset.
This is a minimal public-catalog browsing tool with only 2 read-only tools. Both tools have descriptions and basic schemas, but significant quality gaps prevent a higher score. Tool naming is adequate (verb-noun form), but parameter descriptions are minimal. The 'search_agents' tool lacks field-level documentation of what fields are returned. The 'get_overview' tool has no input parameters but its output schema is completely undocumented. Neither tool includes error handling guidance or recovery paths. The server is STDIO-only, which is a hard architectural limitation that prevents remote accessibility and enterprise adoption.
Get a machine-readable overview of FindAgent — what it is, how to connect, and the key public routes. Read-only, no auth.
Search FindAgent's public catalog of vetted MCP "doer" agents (coding, data analysis, customer support, design, DevOps, and more). Returns matching agents with slug, title, tagline, and description. Read-only, no auth.
Output schemas completely undocumented. Neither tool declares what fields or structure LLMs should expect in responses. 'get_overview' especially returns raw fetched text with no schema guidance.
No error handling or recovery guidance. If 'search_agents' returns no results or a malformed query, the error response does not guide the LLM on what to do next (retry with different terms, check syntax, etc.).
Parameter descriptions are minimal. 'query' in 'search_agents' has a 1-line description with generic examples. No guidance on max length, format constraints, or what constitutes a valid search term.
STDIO-only transport limits production use. This server cannot be accessed by hosted MCP clients or integrated into cloud-native agent platforms. Clients must run it as a local subprocess.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 42 | 2026-07-28+ | v2 |
'get_overview' input schema is an empty object with no documented purpose. The tool description says it returns 'machine-readable overview' but does not specify what fields or structure to expect, or how to use the returned data.
No response field documentation. 'search_agents' description promises 'slug, title, tagline, and description' but this is buried in natural language, not a formal schema. LLMs cannot reliably extract these fields or chain to downstream tools.