Agentty defines 22 tools with basic schemas and descriptions, but falls short of production quality. Most tools have 1-2 sentence descriptions (avg ~40 chars, well below the 194-char baseline) that lack context for LLM selection. Parameter descriptions are minimal or absent. No output schemas are documented. Critical tools like 'bash' and 'write' lack error handling guidance. Tool names are reasonable (verb-noun style) but descriptions do not explain WHEN to use each tool or what happens on failure. Schemas are present but sparse, many parameters lack type constraints or validation hints. No evidence of idempotency, pagination, or composition patterns.

• Expand tool descriptions to 100 - 200 chars. Include WHAT the tool does, WHEN to use it (vs similar tools), and key side effects. E.g., 'bash: Execute shell commands in the workspace. Use for complex operations; prefer grep/glob for simple searches. Commands are irreversible, no undo. Returns stdout, stderr, exit code.'
• Document output schemas for all tools. Specify field names, types, and whether results are paginated. E.g., read() returns {content: string, line_count: int, encoding: string}; search_docs() returns {results: [{id, score, text}], total: int, has_more: bool}.
• Add parameter descriptions explaining format, constraints, and examples. E.g., bash command: 'Shell command to execute (max 10000 chars, no interactive prompts). Returns {stdout, stderr, exit_code}.'
• Implement pagination for list-returning tools. Add limit (default 20, max 100) and offset/cursor parameters. Document result caps in descriptions to prevent context window exhaustion.
• Add error handling guidance to destructive tools. E.g., write() should document: 'Creates or overwrites file. Returns {path, bytes_written, success}. On permission error, returns {error: "permission_denied", path, suggestion: "Check file permissions"}.' Offer dry-run or confirmation for bash/task/skill.