Single tool 'execute_skill_action' has a minimal description (79 chars) that lacks actionable context for LLM selection. Input schema is present but parameter 'query_payload' has only a generic description ('Input execution payload for this skill') with no format, length, or content constraints. No output schema is documented, the tool returns a hardcoded JSON string with no structure definition. No error handling guidance, no idempotency markers, and no indication of what the tool actually does beyond vague 'AI Agent Operations'. The tool name uses 'execute' (generic verb) rather than a specific action verb. No parameter validation rules, no examples of valid payloads, and no recovery guidance if execution fails.

• Document the output schema explicitly. Define what fields the response contains (e.g., status: enum[SUCCESS|FAILURE|TIMEOUT], message: string, result: object, execution_time_ms: number). LLMs need this to plan downstream actions.
• Rewrite the parameter description for 'query_payload' to specify format and constraints: 'A JSON string containing the skill execution request. Must include: {"action": string, "params": object}. Max 10KB. Example: {"action": "click_element", "params": {"selector": "button.submit"}}'
• Expand the tool description to 100-150 chars explaining purpose and use case: 'Execute a GenPark AI skill (e.g., browser automation, data extraction). Pass a JSON payload with action and parameters. Returns execution status, result, and timing. Use when you need deterministic, verified AI operations.'
• Add error handling: return distinct status values (SUCCESS, FAILURE, TIMEOUT, INVALID_PAYLOAD) with actionable error messages. E.g., 'INVALID_PAYLOAD: Missing required field "action". Valid actions: click_element, extract_text, navigate_url.'
• Rename tool to a more specific verb: 'invoke_skill_action' or 'run_skill_action' if it triggers execution, or 'call_genpark_skill' if it's a GenPark-specific operation. Avoid generic 'execute'.