Two tools with clear verb-noun naming (fetch_issue, list_issues) and documented parameters. Descriptions are present but generic (194-char baseline; these average ~180 chars). Parameter schemas are complete with types and defaults, but descriptions lack actionable constraints (e.g., no enum for 'state', 'sort', 'direction'). Output schemas are documented in docstrings but not formally declared in the tool registration. Error handling is minimal, generic exceptions without recovery guidance. No tool annotations (readOnlyHint, idempotentHint). Missing pagination guidance despite list_issues returning up to 100 items.

• Add enum constraints to list_issues parameters: state=['open','closed','all'], sort=['created','updated','comments'], direction=['asc','desc']. Replace free-form strings with formal constraints.
• Enhance error messages with recovery guidance: 'Repository not found. Verify format is owner/repo_name (e.g., torvalds/linux). Call list_issues() to discover available repos.' instead of generic 'Not found'.
• Declare tool annotations in FastMCP registration: both tools are read-only and idempotent. Use readOnlyHint=true and idempotentHint=true in tool metadata.
• Add pagination guidance to list_issues description: 'Returns up to 100 issues per call. For large repositories, filter by state, labels, or date range to reduce result size. No cursor-based pagination available; use sort and direction to navigate results.'
• Document output schema formally in tool registration (not just docstring). Specify that fetch_issue returns {title, body, state, comments, html_url, created_at, updated_at, labels[], is_pull_request} and list_issues returns [{number, title, body, state, comments, html_url, created_at, updated_at, labels[], is_pull_request}].