Two tools with complete schemas and descriptions. pha_user_whoami is minimal but clear (READ_ONLY). pha_user_search has comprehensive parameter coverage (16 params, all typed and described) with pagination support (limit, after, fetch_all). Descriptions are adequate (50-200 chars range). However, output schemas are not documented in the source, LLMs cannot see what fields to expect from responses. No error handling guidance visible. Tool names follow verb_noun pattern but lack clarity on what 'pha_' prefix means. Missing per-parameter constraints (e.g., limit bounds, enum values for 'order' and 'query_key'). No mention of idempotency or retry safety.

• Document output schemas for both tools. Example for pha_user_whoami: {"type": "object", "properties": {"id": {"type": "string"}, "username": {"type": "string"}, "email": {"type": "string"}, ...}}. This enables LLMs to chain results to other tools.
• Convert parameter constraints to JSON Schema enums and bounds. For pha_user_search: order should be enum ["newest", "oldest", "relevance"]; limit should have minimum: 1, maximum: 100; query_key should be enum ["active", "admin", "all", "approval"].
• Add error handling guidance to tool descriptions. E.g., 'If no results match, try broadening name_like or removing filters. If query_key is invalid, valid options are: active, admin, all, approval.'
• Clarify tool name prefix. Rename to 'phorge_user_whoami' and 'phorge_user_search' (or 'conduit_user_*' if Conduit is the canonical name) to improve LLM discoverability and reduce confusion with other Phabricator tools.
• Add tool annotations to schema. Include readOnlyHint: true for both tools (they are read-only). Include idempotentHint: true to signal safe retries. This enables agents to optimize retry logic.