AI-powered changelog generator with MCP server support - works with most providers, online and local models
The server implements 4 tools with clear action verbs and structured schemas. Tool descriptions are present and moderately detailed (71-200 chars). Schemas are well-formed with typed parameters and enums. However, several parameters lack descriptions, error handling guidance is minimal, and output schemas are not documented. The tools show reasonable composition but miss critical LLM-facing details like actionable error messages and dependency hints.
Analyze staged and unstaged changes in working directory
Comprehensive repository analysis including health, commits, and branches
Manage AI providers - list, switch, test, and configure
Generate AI-powered changelog from commits or working directory changes
Output schemas not documented. Tools return results but LLMs cannot see what fields to expect in responses, forcing guessing and fragile downstream tool chains.
Error handling lacks recovery guidance. Tool definitions provide no guidance on what an LLM should do if a call fails (retry? lookup? ask user?). Errors will be opaque stack traces.
Parameter descriptions incomplete. repositoryPath, author, since, tagRange lack explanation of expected format, valid ranges, and how they interact. LLMs will guess and pass invalid values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 51 | 2025-06-18+ | v2 |
| 2026-03-09 | F | 35 | 1.25.3+ | v1 |
No dependency hints between tools. LLMs cannot see that analyze_repository should often precede generate_changelog, or how to chain results. No documentation of which parameters are mutually exclusive.
Destructive/reversible operations (generate_changelog writes files, configure_providers modifies state) lack explicit safety annotations or confirmation patterns. No dry-run option visible.