improve-app
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) because it reads and processes data from documentation files in the project's 'docs/' folder to determine journey state and actions. 1. Ingestion points: The agent reads docs/IMPROVE-APP-PLAN.md, docs/CUSTOMER.md, docs/DESIGN.md, docs/POSITIONING.md, docs/EXPERIMENTS.md, and docs/PRODUCT.md. 2. Boundary markers: Absent. No specific delimiters are used to separate user-provided content in these files from agent instructions. 3. Capability inventory: The skill is scoped to reading and writing project documentation and does not possess high-risk capabilities like direct network exfiltration or unauthorized shell execution. 4. Sanitization: Absent. The skill does not validate or sanitize the content of the documentation files before processing.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to suggest shell commands to the user (npx skills add) for installing missing project dependencies.
- [REMOTE_CODE_EXECUTION]: The suggested dependency installation involves downloading and executing remote code from the author's (wondelai) repository, which is recognized as a vendor-specific resource.
Audit Metadata