Open-source licenses MCP (SPDX). Look up open-source licenses by SPDX id, search the license list, and fetch full license text from the official SPDX license-list-data. Keyless.
Three tools with complete schemas and clear descriptions. All tools follow verb_noun naming (lookup_, search_, get_). Descriptions are concise (50-100 chars) and explain purpose well. Input schemas are properly typed with required fields marked. However, output schemas are not documented, LLMs cannot predict response structure. Error handling is minimal (generic 'not found' messages). No parameter constraints (e.g., limit 1-100 is enforced in code but not declared in schema). Missing idempotency hints and recovery guidance.
Fetch the full license text for an SPDX id (e.g. "MIT"). Keyless.
Look up an open-source license by its exact SPDX id (e.g. "MIT", "Apache-2.0", "GPL-3.0-only", "BSD-3-Clause"). Returns the full name, OSI/FSF approval, deprecation status, and reference URLs. Keyless.
Search the SPDX license list by keyword (matches the id or name), e.g. "GPL", "creative commons", "mozilla". Keyless.
Output schemas not documented. LLMs cannot predict response structure (e.g., fields returned by lookup_license, search_licenses result format). This forces agents to guess and wastes tokens on exploratory calls.
Parameter constraints not declared in schema. 'limit' in search_licenses is clamped to 1-100 in code but schema has no min/max. LLMs cannot see the constraint and may pass invalid values.
Error messages lack recovery guidance. 'No license with that exact SPDX id' does not tell the LLM to call search_licenses. Agents need explicit next-step hints.
No idempotency or side-effect declarations. All tools are read-only (correct), but this is not signaled via tool annotations (readOnlyHint). Agents cannot distinguish safe retry patterns.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
No pagination support in search_licenses response. If SPDX list grows, returning all matches (capped at limit) could still exceed context. No next_cursor or offset mechanism.