This server has critical definition gaps. Of 6 tools, only 3 have visible input schemas (migrate_v1_to_v2, validate_benchmark_suite, check_toolkit_hygiene). The other 3 (fetchJars, integrationTest, check_doc_index) show no input schema in the provided source. Descriptions exist but are inconsistent in quality: some are detailed (migrate_v1_to_v2: 'Create a reviewable Major Feature Contract v2 draft...'), others are vague (integrationTest: 'Runs the witness agent integration harness'). No tool has documented output schemas. No error handling guidance. No parameter validation rules. Tool names are reasonable (verb-noun style) but lack the precision needed for LLM disambiguation. This is a development/CI toolkit, not a user-facing agent tool, which explains the gaps, but MCP servers must still meet the rubric standard.

• Add explicit input schemas to all 6 tools. For fetchJars and integrationTest (Gradle tasks), define parameters even if empty (e.g., 'no parameters required'). For check_doc_index, add optional 'project_root' parameter with type Path and description.
• Document output schemas for all tools. E.g., migrate_v1_to_v2 should return {success: boolean, output_path: string, migration_notes: string[], warnings: string[]}. validate_benchmark_suite should return {valid: boolean, summary: object, errors: string[]}.
• Add error handling guidance to each tool description. E.g., 'If validation fails, check that the input file matches the expected JSON schema. If the file is missing, ensure the path is absolute or relative to the project root.'
• Expand parameter descriptions with format/constraint hints. E.g., 'source_path: Path to v1 contract JSON file (must exist, must be valid JSON matching v1 schema, typically in contracts/ directory).'
• Rewrite tool descriptions to include WHEN/WHY. E.g., 'integrationTest: Validates the witness agent by running integration tests against fixture JVMs. Call this after building the agent jar to verify runtime behavior instrumentation works correctly.'