Two tools with minimal descriptions and incomplete parameter documentation. add_temporal_constraint has a 79-char description (acceptable baseline) but parameters lack descriptions entirely, only type and name are present. solve_stn has a 62-char description and zero parameters, making it impossible for an LLM to understand what it returns or when to call it. No output schemas documented. No error handling guidance. No parameter constraints (enums, ranges, formats). Naming is verb-based (add_, solve_) but descriptions do not explain WHEN to use each tool or what the output structure is.

• Add descriptions to all four parameters of add_temporal_constraint: 'source: Identifier of the source timepoint (e.g., "task_A_start"). Must match a previously added timepoint or be created implicitly.', 'target: Identifier of the target timepoint.', 'min_duration: Minimum duration constraint in seconds (must be >= 0).', 'max_duration: Maximum duration constraint in seconds (must be >= min_duration).'
• Document the output schema for solve_stn in its description: 'Returns {is_consistent: bool, schedule: {[timepoint]: {earliest_time: number, latest_time: number, slack: number}}, node_count: int, error?: string}. If inconsistent, schedule is null and error explains the negative cycle.'
• Add input validation and error messages: 'If min_duration > max_duration, return error: "Invalid constraint: min_duration (X) must be <= max_duration (Y)". If timepoint names are empty strings, return error: "Timepoint names cannot be empty."'
• Expand add_temporal_constraint description from 79 chars to ~150 chars: 'Add an interval constraint between two timepoints: target - source must be in [min_duration, max_duration]. Use this to model task durations, precedence, and deadlines. Timepoints are created implicitly on first reference.'
• Expand solve_stn description to explain when to call it: 'Check if all temporal constraints are consistent (no negative cycles) and compute earliest/latest start times for each timepoint. Call this after adding all constraints to validate the schedule and extract bounds for planning.'