Loading ToolBench
MCP server for the Transkribus REST API — manage collections, documents, recognition, models, and more, plus the Metagrapho Processing API. 304 tools across 23 resource domains.
6 tools with basic schemas and descriptions, but significant gaps. All tools have descriptions (10-50 chars, below the 50-200 char optimum), and input schemas are present but minimal. Parameters lack detailed descriptions explaining constraints, formats, or dependencies. No output schemas documented. Error handling is absent, tools return raw API responses without recovery guidance. Tool names follow verb_noun pattern (good), but descriptions are too terse to guide LLM selection. No parameter validation hints, no enum constraints, no idempotent/destructive markers in descriptions. Composition is reasonable (single responsibility per tool), but chaining support is unclear, output fields not documented.
List collections accessible to the authenticated user
Get the status of a Metagrapho processing job
Create an upload structure for document ingestion
Delete an upload structure
Get the status of an upload
Upload a page image to an upload structure
Descriptions are too short (10-50 chars) to guide LLM tool selection. Baseline is 50-200 chars. E.g., 'List collections accessible to the authenticated user' lacks context on when to call this vs other discovery tools, what structure is returned, or prerequisites.
Parameters lack descriptions. 'collId' (integer), 'title' (string), 'pages' (array) have type hints but no guidance on format, constraints, or valid ranges. LLMs cannot infer whether collId is 1-10000, what title length is acceptable, or what structure pages must have.
No output schemas documented. Tools return raw API responses (uploadId, pageList, jobId, etc.) but LLMs have no schema to understand what fields to expect or how to chain results to downstream tools. Baseline: 100% of A+ tools document return types.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2025-06-18+ | v2 |
No error handling guidance. Tools do not document what errors are retryable, user-fixable, or fatal. No recovery hints (e.g., 'If upload fails, call transkribus_upload_delete to clean up'). Agents cannot self-correct on failure.
Destructive tool (transkribus_upload_delete) lacks confirmation or dry-run support. No description warning that deletion is irreversible. Agents may accidentally delete uploads without safeguards.