analyze-canvas-performance

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The prompt contains conflicting/overriding instructions — an early CRITICAL rule forbids modifying any .yaml/.pa.yaml files during the skill execution, but a later "Offer to Apply Fixes" section instructs the agent to edit those same files and call compile/validate if the user approves, which attempts to override the earlier prohibition and expands the skill's scope beyond the stated audit-only purpose.

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires reading .pa.yaml files and embedding exact formula/property snippets (Code Found, query code, App.Formulas, connection strings, etc.) into the generated HTML/report, which may contain API keys, connection strings, or other secrets — forcing the LLM to handle and output secrets verbatim.