HashPilot has 11 tools with basic schemas and descriptions, but significant gaps prevent higher scores. All tools have descriptions (10-100 chars, mostly adequate), and input schemas are present with type definitions. However, output schemas are entirely undocumented, LLMs cannot plan downstream calls or extract results. Parameter descriptions are minimal (5-20 chars), violating the baseline of 72 chars average. No error handling guidance, no recovery paths, and no actionable error messages. WRITE tools (create_account, transfer_hbar, deploy_contract, create_token) lack confirmation/dry-run patterns despite irreversible consequences. Tool names are clear (verb_noun style), but composition is weak: create_account and transfer_hbar should accept human-readable identifiers (names, emails) instead of opaque IDs. No pagination support for list-like tools (docs_search, network_nodes). Security: no visible secret injection for Hedera SDK credentials. Overall, this is a functional but immature tool set suitable for exploration, not production agents.

• Document output schemas for all 11 tools. Specify field names, types, and whether results are paginated. E.g., docs_search should return [{id, title, content, relevance_score}] with total_count and next_cursor.
• Expand parameter descriptions to 50-150 chars. Add enums for 'network' (mainnet|testnet|previewnet). Add format hints for 'publicKey' (hex string, 66 chars), 'bytecode' (hex encoded), 'constructorParams' (ABI encoded hex).
• Add confirmation/dry-run pattern to WRITE tools. E.g., create_account with dry_run=true returns estimated fees without executing; transfer_hbar with confirm_hash returns a hash for user approval before broadcasting.
• Implement error recovery guidance. E.g., 'Invalid account ID: must be in format 0.0.X. Try search_accounts(name) to find valid IDs.' Return available alternatives when resources not found.
• Support human-readable identifiers in WRITE tools. Accept account_name, account_alias alongside account_id. Resolve names to IDs server-side. Reduces lookup calls and matches user intent in chat.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry