Loading ToolBench
An AI agent that continuously searches for novel AI agent architectures, extracts and evaluates them, and produces diagrams or documentation via MCP tools
Scout7 exposes 8 tools with basic descriptions and partial schemas. Most tools lack comprehensive parameter documentation and output schema definitions. Naming follows verb_noun convention (good), but descriptions are terse (avg ~60 chars, well below the 194-char baseline). Parameter types are present but descriptions are minimal or missing. No error handling guidance, no output schemas documented, and no tool annotations. The server acts as a mesh client orchestrating external services (ollama, searxng, arch7, mem7) rather than implementing tools directly, tool definitions are inferred from code rather than explicitly registered.
Generates an Excalidraw diagram via arch7 from extracted architecture components and connections
Retrieves the text content of a URL via agent-mesh fetch
Writes markdown or JSON content to a file via a file-writing MCP tool (configurable tool name)
Returns the names of architectures already stored in mem7
Searches mem7 to check if a URL has already been processed
Saves a processed article to mem7
Tool definitions inferred from code, not explicitly registered. No visible MCP tool registration with schemas in the codebase.
Descriptions are terse (avg 60 chars vs 194-char baseline). Examples: 'Sends messages to ollama via agent-mesh and returns the response text' (67 chars) lacks WHEN to use, prerequisites, or output structure.
No output schemas documented. LLMs cannot plan downstream calls or extract fields. E.g., ollama.chat returns ChatResponse with nested content array, not documented in tool definition.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
Sends messages to ollama via agent-mesh and returns the response text
Queries searxng via agent-mesh and returns search results
Parameter descriptions missing or minimal. 'query' in memory_search lacks format/length guidance. 'messages' in ollama.chat lacks structure (role/content fields).
No error handling guidance. No recovery hints (e.g., 'If model not found, call list_models()'), no error classification (retryable vs fatal), no actionable error messages.