Loading ToolBench
Local agent session query API - read-only service for browsing and querying agent sessions from Claude, Codex, Gemini, Grok, and other AI agents
Server provides 11 read-only tools with consistent naming (verb_noun pattern: list_, get_, search_, export_) and reasonable descriptions (avg ~120 chars). All tools are READ_ONLY, reducing risk. However, input schemas are partially visible but lack formal JSON Schema type declarations in the provided code excerpt. Parameter descriptions exist but are brief (10-30 chars). Output schemas are not documented. No error handling guidance or recovery patterns evident. Tool composition is sound (each does one thing), but schema completeness and error guidance are weak.
Export a session in various formats (markdown, JSON, JSONL, HTML)
Retrieve the final result or status of a session
Retrieve messages from a session with optional filtering by limit, offset, and direction
Retrieve metadata for a specific session by pattern matching across sources
Retrieve a markdown brief of a session or specific round with summary of asks, tool usage, and outcomes
Retrieve the rounds index for a session - segments of user asks and assistant responses
Check server health and list available sources with any warnings
Output schemas not documented. Tools return data but LLMs cannot plan downstream calls without knowing response structure (fields, types, pagination). Violates pattern:tool and pattern:response-shaper.
Input schemas lack formal JSON Schema type declarations. Code shows parameter names and descriptions but no explicit 'type' fields (string, integer, boolean, array).
No error handling guidance. Tools lack recovery hints (e.g., 'Session not found. Try search_sessions() with a partial ID.'). Violates pattern:recovery-guide.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
List all available API endpoints with their authentication requirements
List all projects (grouped by working directory) with session counts and source information
List all agent sessions across enabled sources with metadata including session ID, source, status, and update time
Search session content by keyword across all sessions and sources
Parameter descriptions are minimal (10-30 chars). 'Session ID or pattern to match' is vague, does it accept regex, glob, or substring? Format constraints missing. Violates pattern:tool-description baseline (72 chars avg).
Pagination not documented. list_sessions, list_projects, search_sessions likely return large result sets but no limit/offset/cursor parameters visible. Violates pattern:paginated-result.