Loading ToolBench
MCP server that integrates with Pieces OS to provide long-term memory capabilities via chat
Single tool 'PiecesLTM' with minimal quality baseline. Tool has a schema and description present, but lacks depth in parameter documentation, output schema definition, and error handling guidance. The description is extremely terse (28 characters), below the 50-200 character target for LLM-optimized tools. Parameter schema is present with types and basic description, but no guidance on expected format, constraints, or dependencies. No documented output schema, pagination support, or error recovery paths. Tool performs a single well-defined action (query LTM), but lacks the structural rigor expected for production-grade agent tools.
Ask Pieces LTM a question
Description is critically short (28 chars). Must be 50-200 chars for LLM selection and context. Current text 'Ask Pieces LTM a question' lacks WHEN to use it, WHAT it returns, or any prerequisites.
No output schema documented. LLMs cannot plan downstream calls or extract structured data. Must specify: return type (string vs object), expected fields, format (markdown/plaintext/JSON), and whether results are paginated or bounded.
Parameter 'question' has no format constraints, length limits, or usage examples. Description is only 27 characters. LLMs need guidance on question complexity, length boundaries, and any topics the LTM cannot answer.
No error handling or recovery guidance. Tool calls external Pieces OS client and chat service with no documented timeouts, retry behavior, or error categorization. LLMs receive raw errors with no actionable next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 42 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 8 | - | v1 |
Tool name 'PiecesLTM' lacks action verb prefix. Should be 'query_pieces_ltm' or 'ask_pieces_ltm' to clearly signal intent. LLMs parse intent from verb prefixes (get_, create_, search_, etc.) and rely on this convention to disambiguate tool selection.
Tool description does not indicate state modification. Code shows tool queries an external LLM service with memory retention (7-day LTM), which has side effects, but description lacks any note that 'This updates Pieces LTM memory' or marks it as a command.