Server defines 6 tools with complete input schemas and descriptions. Naming follows verb_noun convention (ossinsight_*). Descriptions are adequate (50-100 chars) but lack actionable context for LLM selection. All parameters have type definitions and descriptions. No output schemas documented. Error handling is basic (retry logic present but no recovery guidance in tool descriptions). Missing tool annotations (readOnlyHint, idempotentHint). Parameter descriptions lack constraints (e.g., enum values for 'period' and 'metric' are stated in description text, not as JSON Schema enums).

• Convert enum constraints to JSON Schema. Replace 'period': {'type': 'string', 'description': 'past_24_hours | past_week...'} with 'period': {'type': 'string', 'enum': ['past_24_hours', 'past_week', 'past_month', 'past_3_months'], 'default': 'past_week'}. Same for 'metric' and 'sort'.
• Document output schemas for each tool. Add a 'Returns' section to descriptions: 'Returns object with fields: owner (string), repo (string), stars (integer), forks (integer), language (string), license (string), description (string).'
• Expand tool descriptions to 100-150 chars with actionable context. E.g., 'Get GitHub repository metadata (stars, forks, language, license). Use this to inspect a specific repo; use ossinsight_search to find repos by keyword.'
• Add tool annotations via MCP SDK. Mark all tools with readOnlyHint=true and idempotentHint=true so agents know they are safe to retry without side effects.
• Include error recovery guidance in descriptions. E.g., 'If repo not found, try ossinsight_search() with the repo name. If rate-limited, retry after 60 seconds.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry