Phero exposes 2 tools with minimal documentation. Both tools have basic descriptions (60-80 chars) and input schemas with typed parameters, but lack depth in parameter descriptions and output schema documentation. The 'agent' tool creates runtime agents (appropriate verb), and 'skill' tool executes skills (appropriate verb), but descriptions are generic and do not explain WHEN to use each tool or what distinguishes them. Parameter descriptions are present but terse. No output schemas are documented. Error handling guidance is absent. Naming is acceptable but descriptions fall short of the 50-200 char LLM-optimized baseline for A-grade tools.

• Document output schemas for both tools. Specify: agent returns {agent_id, status, result, execution_time}; skill returns {skill_name, base_path, instructions, available_commands}. Include field types and descriptions.
• Expand parameter descriptions to 50-100 chars. E.g., 'name': 'Unique identifier for the agent (alphanumeric, 3-50 chars). Used to reference the agent in subsequent calls.' instead of 'The name of the agent to create and run.'
• Add WHEN-to-use guidance in tool descriptions. E.g., agent: 'Delegate complex multi-step tasks to a specialized agent. Use when the task requires a distinct role or expertise.' skill: 'Load and execute a pre-built skill within the current conversation. Use for single-purpose utilities that don't require a separate agent context.'
• Document error scenarios and recovery paths. E.g., 'If agent creation fails due to invalid name, try a simpler name (letters/numbers only). If skill not found, call list_available_skills() first.'
• Add constraints to parameters. E.g., 'name' enum or pattern (alphanumeric, 3-50 chars); 'command' with available skill names or reference to discovery tool.

Inferred effective spec: 2026-07-28+.

• Relies on HTTP+SSE-only transport (deprecated) - migrate to Streamable HTTP