Delega CLI exposes 16 tools with basic schemas and descriptions. All tools have input schemas with typed parameters and descriptions present. However, descriptions are often terse (10-50 chars), parameter descriptions lack actionable constraints (enums, ranges, formats), and output schemas are not documented. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. Error handling is minimal, no recovery guidance or actionable error messages visible in source. Tool naming follows verb_noun convention (agents list, agents create, human register) but some names are ambiguous (human status vs human result both read-only, unclear distinction). No evidence of pagination support for list tools. Security: login tool accepts no params (good, no credential exposure), but no audit trail or permission-gate patterns visible.

• Expand all descriptions to 50-200 characters. Include WHAT the tool does, WHEN to use it, and what it returns. E.g., 'agents list' → 'List all agents in your fleet. Returns agent IDs, names, roles, and status. Call this first to discover available agents before assigning tasks.'
• Document output schemas for all tools. Specify field names, types, and whether results are paginated. E.g., agents list returns {agents: [{id, name, display_name, role, active, created_at}], total: number}.
• Add enum constraints to parameters with known valid values. E.g., role: enum ['worker', 'coordinator', 'admin']; rule: enum ['daily', 'weekly', 'monthly', 'yearly']. Replace free-form strings with formal constraints.
• Add min/max bounds to numeric parameters. E.g., timeout-seconds: 60 - 1200; interval: 1 - 365; priority: 1 - 4. Include these in parameter descriptions and schema.
• Add tool annotations to all tools. Mark agents delete and recurring delete with destructiveHint=true; mark agents list, human status, human result with readOnlyHint=true. This enables agents to reason about side effects and retry safety.