MCP Server for Azure AI Foundry Claude Integration with Software Engineer Agent and GitHub integration
Server provides 8 tools with basic schemas and descriptions, but quality is inconsistent. All tools have descriptions (10-80 chars, mostly adequate) and input schemas with type definitions. However, parameter descriptions are sparse or missing context. No output schemas documented. Error handling guidance absent. Tool naming follows verb_noun convention but lacks specificity in some cases (e.g., 'analyze' is generic; 'chat' and 'software_engineer_chat' are near-duplicates). No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all being read-only. Composition issues: 'chat' and 'software_engineer_chat' perform similar functions, forcing LLM to reason about which to use. GitHub tools lack pagination/limit parameters despite potentially returning large result sets. No documented output structure means agents cannot plan downstream calls reliably.
Analyze text or code with Claude Opus 4.5
Send a message to Claude Opus 4.5 and get a response
Analyze a GitHub repository with Software Engineer Agent
Get content of a file from a GitHub repository
Get file structure of a GitHub repository
List all GitHub repositories from leonardo-matheus
Review a specific file from a GitHub repository
No output schemas documented. LLMs cannot infer what fields to expect from tool responses, forcing them to guess at downstream field names and breaking tool chaining.
Duplicate/overlapping tools: 'chat' and 'software_engineer_chat' both send messages to Claude variants. LLM must reason about which to use, wasting tokens and risking wrong selection.
GitHub list/get tools lack pagination (limit, offset, page_size) and result-count documentation. Large repos could return unbounded results, exhausting context window.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | <=2025-11-25 | v2 |
Chat with Software Engineer Agent that has access to GitHub repos
Parameter descriptions missing or minimal. 'repo' param in github_analyze_repo lacks format guidance (e.g., 'owner/repo' vs 'repo-name'). 'analysis_type' enum is present but lacks guidance on when to use each type.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 8 tools are read-only but this is not declared, forcing LLMs to infer safety from descriptions.