Server has 7 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-60 chars, mostly short), and input schemas are present with typed parameters. However, descriptions are often vague or lack context for LLM selection. No output schemas documented. Error handling is absent, no guidance on retryability, user-fixable errors, or recovery paths. No tool annotations (readOnlyHint/destructiveHint). Parameter descriptions are minimal. The 'rationale' parameter in buy_shares/sell_shares is unusual but undocumented. No pagination, no result limits documented. Security: credentials (PUSHOVER_USER, PUSHOVER_TOKEN) are injected server-side (good), but no audit logging visible. Composition is reasonable, tools are single-purpose, but output chaining not verified.

• Add output schemas to all tools. Document return types as structured objects with typed fields. E.g., get_balance should return {balance: float, currency: string, last_updated: ISO8601}.
• Expand tool descriptions to 100-200 chars. Include WHAT the tool does, WHEN to use it (vs similar tools), and WHAT it returns. E.g., 'Get the current cash balance (not including holdings) for an account. Use this to check available funds before buying shares. Returns balance in USD.'
• Add error handling guidance. Document failure modes and recovery paths. E.g., 'buy_shares may fail if insufficient balance, check get_balance first. If symbol not found, use lookup_share_price to verify.'
• Implement tool annotations in FastMCP. Mark get_balance/get_holdings/lookup_share_price with readOnlyHint=true. Mark buy_shares/sell_shares/change_strategy/push with destructiveHint=true. Mark buy_shares/sell_shares with idempotentHint=false (retries risk duplicate trades).
• Document the 'rationale' parameter. Explain why it's required, how it's used (audit trail? strategy validation?), and what format is expected. If optional, make it so and provide a default.