Agent Fleet exposes 8 tools with minimal documentation. All tools have basic descriptions (10-50 chars), but lack parameter-level descriptions entirely. Input schemas are present but sparse: most parameters have only 'type' and 'description' at the parameter level, with no constraints, enums, or validation guidance. No output schemas are documented. Error handling is absent, no recovery guidance, categorization, or actionable error messages. Tools like Bash and Edit are high-risk (WRITE operations) with no confirmation, dry-run, or permission-gate patterns. Naming is acceptable (verb-noun style: Read, Glob, Grep, Bash, Edit, Write, WebSearch, WebFetch) but descriptions are too terse to guide LLM selection. No tool composition guidance, no idempotency markers, and no security patterns (e.g., secret injection, audit trails). This is a minimal MCP wrapper around agent capabilities, not a production-grade tool suite.

• Add 50-150 character descriptions to each tool explaining WHAT it does, WHEN to use it, and any prerequisites. E.g., 'Read file contents from disk. Use for source code, config files, or documentation. Requires valid file path; returns raw file text.'
• Document every parameter with type, format, and constraints. E.g., 'path (string): Absolute or relative file path. Must exist and be readable. Supports glob patterns (*, **, ?) for Glob tool only.'
• Define output schemas for all tools. E.g., Read returns {content: string, size: number, encoding: string}; Glob returns {matches: string[], count: number, pattern: string}.
• Add confirmation/dry-run patterns to Bash, Edit, Write. E.g., 'bash_dry_run' tool that shows what would execute without side effects, or require explicit 'confirm=true' parameter for destructive operations.
• Implement error categorization and recovery guidance. E.g., 'File not found: /path/to/file. Use Glob to search for similar files.' or 'Invalid bash syntax: unexpected token. Check command for typos or quote mismatches.'