Server has 4 tools with clear verb-noun naming (execute_target, get_output, search_output, list_targets). All tools have descriptions (34-72 chars, within baseline range). Input schemas are present with type definitions and parameter descriptions. However, output schemas are not documented, LLMs cannot predict return structure. Error handling is minimal; no recovery guidance or actionable error messages. Missing tool annotations (destructiveHint for execute_target, readOnlyHint for others). Parameter descriptions are adequate but lack format constraints and validation rules. No pagination support for search_output despite returning potentially large result sets.

• Document output schemas for all tools. Example: execute_target returns {execution_id: int, target: string, stdout: string, stderr: string, exit_code: int, timestamp: float}. This enables LLMs to chain tools and extract required fields.
• Add tool annotations via FastMCP. Mark execute_target with destructiveHint=true. Mark get_output, search_output, list_targets with readOnlyHint=true. This signals safety to agents without requiring description parsing.
• Add pagination to search_output. Accept offset and limit parameters (default limit=20, max=100). Return total_matches count so agents know if results are truncated and can page through all matches.
• Enhance error messages for execute_target. When exit_code != 0, return structured error: {error: 'make target failed', target: string, exit_code: int, stderr: string, suggestion: 'Check target syntax or dependencies'}. This guides agent recovery.
• Add format constraints to parameter descriptions. Example: 'target: The name of a make target (alphanumeric, underscore, hyphen; max 64 chars)'. 'pattern: A valid Python regex pattern (max 256 chars)'. This prevents invalid input from LLMs.