environments
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill provides guidelines to prevent hardcoded secrets, requiring that they be kept out of repositories and only accessed through environment variables or secure operator files.
- [DATA_EXFILTRATION]: Strict environment isolation is enforced, ensuring that development environments never use production keys or access production databases, effectively preventing accidental data exposure.
- [COMMAND_EXECUTION]: The skill uses official CLI tools from trusted vendors for configuration and deployment tasks, emphasizing the use of scoped tokens and time-boxed sessions to maintain security.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external project-specific instructions and mapping records. (Ingestion points: SKILL.md, environments-and-branches.md; Boundary markers: Absent for external prose; Capability inventory: Execution of CLI tools via npx and cloudflared, and modification of local .env configuration files; Sanitization: Configuration variables are validated against a strict Zod schema at build and boot time).
- [DATA_EXFILTRATION]: Guidance for using Cloudflare quick tunnels includes explicit warnings against exposing production data and instructions to limit access to authorized users only.
Audit Metadata