Loading ToolBench
SDK and integrations for MemoryRouter, a service that enables AI systems to remember conversation history. Includes Open WebUI integration for uploading conversation history and patches for OpenClaw plugin APIs.
Single tool 'action' has critical definition gaps. Tool name lacks action verb (should be 'upload_conversation_history'). Description is adequate (145 chars) but parameters are severely underdocumented. Input schema shows three parameters (body, __user__, __event_emitter__, __event_call__) but lacks type clarity and parameter descriptions. The 'body' parameter is a generic dict with no schema detail. No output schema documented. No error handling guidance. Tool appears to be Open WebUI-specific integration code, not a standard MCP server tool definition.
One-time upload of all conversation history to MemoryRouter. Uploads Open WebUI conversation history to MemoryRouter so the AI remembers everything from day one. After this, MemoryRouter's proxy handles new conversations automatically.
Tool name 'action' lacks action verb. Should be 'upload_conversation_history' or 'upload_to_memoryrouter' to signal intent to LLM.
Input parameter 'body' is typed as generic dict with no schema detail. LLM cannot infer what fields are required or their types. Must document expected structure (e.g. conversation_id, messages array, user_id).
Parameters __user__, __event_emitter__, __event_call__ are undocumented. Descriptions missing entirely. LLM cannot determine when/how to pass these. Mark as optional and explain purpose in descriptions.
No output schema documented. Tool returns Optional[dict] but LLM has no visibility into response structure. Must document success/failure response shapes.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
No error handling guidance. Code shows validation failures (missing API key, already uploaded) but tool description does not explain recovery steps or error conditions to LLM.