Loading ToolBench
Server has 5 tools with consistent naming (verb_noun pattern: get_*) and reasonable descriptions (100-250 chars). All tools are READ_ONLY and well-scoped. However, input schemas lack proper type constraints (no enums for date_str despite fixed format), parameter descriptions are minimal (e.g., 'Number of days to look back' lacks min/max bounds), and output schemas are completely undocumented. Error handling returns plain strings rather than structured guidance. No tool annotations (readOnlyHint, etc.) despite all being read-only. Descriptions are functional but not LLM-optimized per pattern:tool-description baseline (target 50-200 chars; these are 100-250).
Get top activities/applications by time spent. Args: date_str: Date to query - 'today', 'yesterday', or 'YYYY-MM-DD' limit: Maximum number of activities to show (default: 10) Shows which specific applications and websites you spent time on, ranked by duration. Includes productivity classification for each.
Get time spent by category. Args: date_str: Date to query - 'today', 'yesterday', or 'YYYY-MM-DD' Shows high-level categories like Software Development, Communication, Reference & Learning, etc. with time and productivity classification.
Get productivity breakdown by hour. Args: date_str: Date to query - 'today', 'yesterday', or 'YYYY-MM-DD' Shows when during the day you were most/least productive. Useful for identifying peak productivity hours and scheduling deep work.
Get productivity pulse trend for the last N days. Args: days: Number of days to look back (default: 7, max 14) Shows the daily productivity pulse with visual bars and calculates averages. Useful for identifying patterns and trends over time.
Output schemas completely undocumented. All 5 tools return plain strings with no structured schema definition. LLMs cannot parse results reliably or chain downstream calls.
Parameter 'days' in get_productivity_trend lacks min/max bounds. Description says 'max 14' but no JSON Schema constraint enforces it. LLMs may pass invalid values.
Parameter 'date_str' accepts 'today', 'yesterday', or 'YYYY-MM-DD' but no enum constraint. Free-form string invites hallucinated dates like '2024-13-45'.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
Get today's complete RescueTime productivity summary. This is the recommended daily check-in tool. Returns: - Productivity pulse (0-100 score) - Total time logged - Time breakdown by productivity level - Productive vs distracting percentages
Error handling returns plain strings ('Authentication error: ...', 'API error: ...'). No recovery guidance or error classification. LLM cannot determine if error is retryable or user-fixable.
No tool annotations despite all tools being read-only. Missing readOnlyHint would help LLMs understand these are safe to call without confirmation.