A second set of eyes on your system prompts.

Paste a prompt and get structured feedback on clarity, injection surface, ambiguity, and failure modes — the review a senior engineer would give, minus the wait. Scores are heuristic and reproducible: no model call.

• ✓Role & objective clarity
• ✓Sectioned structure (Role / Instructions / Output / Examples)
• ✓Ambiguity phrase detection
• ✓Instruction priority & trust boundaries
• ✓Delimiter / untrusted-data framing
• ✓Secret & sensitive-data hygiene
• ✓Output format constraints
• ✓Failure-mode enumeration
• ✓Tool scoping (when tools provided)
• ✓Concrete rewrite suggestions