Loading ToolBench
Generates creative ad variations and storyboards for social media content with hook intensity scoring and CTR estimation
Single tool 'execute_skill_action' has a reasonable name starting with verb, but critical gaps undermine quality. Description is 95 chars (below 10-char minimum for actionable guidance). Input schema present with two params, both typed as strings with descriptions, but lacks enums, ranges, or validation constraints. No output schema documented, LLM cannot predict return structure (storyboard_scenes, estimated_ctr_pct, hook_intensity_score visible only in client.py, not in tool definition). No error handling, recovery guidance, or idempotency markers. Overall definition quality is poor.
Execute the generative social creative ad variation synthesizer skill
No output schema documented. LLM cannot predict return structure (storyboard_scenes array, estimated_ctr_pct, hook_intensity_score). Violates pattern:tool requirement that output schema be documented.
Tool description is 95 characters, below actionable minimum. Does not explain WHEN to use this tool, WHAT it returns, or any prerequisites. Fails pattern:tool-description guidance.
Input parameters lack constraints. 'target_aspect_ratio' defaults to '9:16' but no enum or pattern validates allowed values (9:16, 16:9, 1:1, etc.). LLM can hallucinate invalid ratios.
No error handling or recovery guidance. If synthesis fails, LLM receives no actionable error message or retry strategy. Violates pattern:recovery-guide.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 34 | 2026-07-28+ | v2 |
Tool definition inferred from mcp_server.py JSON output rather than explicit schema registration.