Loading ToolBench
VS Code Copilot MCP server and hooks for Claude-Mem project memory.
Server defines 12 tools with complete JSON schemas and descriptions. Naming follows verb_noun convention (cmem_search, cmem_remember, cmem_status). Descriptions are present but brief (avg 60 chars, below 194-char baseline). All tools have input schemas with type definitions and required fields. However, parameter descriptions are sparse or missing entirely, most params lack guidance on format, constraints, or when to use them. Output schemas are not documented. Error handling is minimal; no recovery guidance or actionable error messages visible in handler code. Security: no secrets in params (good), but no audit logging or permission gates evident.
Mark a logical Claude-Mem session complete.
Fetch full Claude-Mem observations by id.
List projects known by the Claude-Mem worker.
Get recent compact Claude-Mem context for a project.
Record a concise memory note in Claude-Mem.
Record an implementation change, files, commands, tests, and result in Claude-Mem.
Record a project decision and rationale in Claude-Mem.
Parameter descriptions missing or minimal. Most params (project, limit, offset, dateStart, dateEnd, orderBy, depth_before, depth_after, etc.) lack guidance on format, constraints, or when to use them. LLMs cannot infer whether 'limit' accepts 1-100 or 1-1000, or what 'orderBy' values are valid.
Output schemas not documented. Tool descriptions do not specify what fields are returned or their types. Agents cannot plan downstream calls or extract data reliably without knowing response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2026-07-28+ | v2 |
Search Claude-Mem project memory.
Create or reuse a logical Claude-Mem session for VS Code Copilot.
Check Claude-Mem worker health and bridge configuration.
Ask Claude-Mem to summarize a logical session.
Fetch memory around an anchor observation or query.
Error handling lacks recovery guidance. Handler code (src/tools/handler.ts not shown) likely returns raw errors without actionable next steps. No evidence of error categorization (retryable vs user-fixable vs fatal) or suggestions for recovery.
Tool descriptions too brief (avg 60 chars vs 194-char baseline). Descriptions like 'Search Claude-Mem project memory' and 'Record a concise memory note' lack context on WHEN to use the tool, what it returns, or how it differs from similar tools (e.g., cmem_remember vs cmem_remember_decision vs cmem_remember_change).
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Risk field is present in metadata but not exposed in tool definitions. Agents cannot distinguish safe read-only tools from destructive writes without explicit annotations.