deployment-plan
Fail
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill instructs the agent to interview the user to extract sensitive information, including environment variables, secret generation methods, and admin/bootstrap credentials. The 'Writing rules' in
SKILL.mdexplicitly mandate that 'Every command must be copy-paste-runnable... using its real... env vars' and 'No placeholders where a real value is known.' This leads to the storage of plaintext secrets and credentials within the generateddocs/deployment-plan/files. - [INDIRECT_PROMPT_INJECTION]: The skill has a large ingestion surface for untrusted data, which could lead to malicious instructions influencing the generated runbook.
- Ingestion points: The agent is instructed to read
docker-compose*.yml,Dockerfile,.github/workflows/,deploy/scripts, nginx configs,.env.example, migration setups, and various Markdown files from the repository. - Boundary markers: The instructions do not define delimiters or warnings to ignore instructions embedded within the ingested files.
- Capability inventory: The skill has the capability to write multiple Markdown files to the local file system (
docs/deployment-plan/). - Sanitization: There is no evidence of sanitization or validation performed on the data read from the repository files before it is used to generate the deployment plan.
Recommendations
- AI detected serious security threats
Audit Metadata