setup-deploy
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from project configuration files and CLI tool outputs to suggest deployment settings, creating a potential surface for indirect injection if these sources were compromised.
- Ingestion points:
CLAUDE.md, platform-specific configuration files (e.g.,fly.toml,render.yaml), and command-line outputs fromfly,vercel, andgh. - Boundary markers: Employs the
AskUserQuestiontool to verify and confirm all inferred values with the user before persistence. - Capability inventory: Utilizes
Bashfor discovery andWrite/Editfor updatingCLAUDE.md. - Sanitization: Instructions explicitly forbid printing secret tokens and mask credentials during verification steps.
- [COMMAND_EXECUTION]: The skill executes platform-specific CLI tools and shell utilities to discover deployment metadata and verify health endpoints. These operations are essential to its primary function and follow the principle of least privilege.
- [EXTERNAL_DOWNLOADS]: The skill documentation directs users to
aside.comto obtain a browser tool for automated dashboard interactions. The agent is strictly prohibited from performing the download or installation itself, ensuring user control over the local environment and preventing automated remote code execution.
Audit Metadata