Loading ToolBench
Repository tooling and recipe catalog for Goose with MCP extension definitions and project skills. Not an installable Python library.
Single tool 'list_pending_diary_analyses' has a reasonable description (89 chars) and input schema with three parameters (limit, include_stale, tenant), each with type and description. However, the tool is defined in a bash script (scripts/analyse-queue.sh), not in Python MCP server code. The server appears to be a shell wrapper around Kamerplanter API calls via curl, not a proper MCP server implementation. No evidence of MCP server framework, tool registration mechanism, or structured output schema. The tool definition is inferred from bash argument parsing, not explicitly registered.
List pending diary analyses from the Kamerplanter queue with optional filtering by tenant and limit
Tool definition inferred from bash script, not explicitly registered in MCP server code. No Python MCP framework detected.
No output schema documented. Response structure is inferred from Python JSON parsing in bash, but fields, types, and structure are not declared for the LLM.
No pagination support visible. Tool accepts 'limit' parameter but no 'offset', 'cursor', or 'total_count' in response. Large result sets will blow context.
Error handling is implicit (bash 'set -e' and curl status checks). No structured error responses, recovery guidance, or actionable error messages for the LLM.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 31 | 2025-06-18+ | v2 |
API key (KAMERPLANTER_API_KEY) is passed as environment variable, not as tool parameter. However, no evidence of secure secret injection pattern or audit logging.