env
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructs the agent to read sensitive local files, specifically
.kamal/secretsand.kamal/secrets-common, in order to identify existing secret names. While this is necessary for the skill's functionality as a configuration assistant, it exposes the contents of these credential files to the agent's context. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted content from the project's configuration files.
- Ingestion points: The agent is directed to read
config/deploy.ymland.kamal/secretsin theSKILL.md'Before You Start' section. - Boundary markers: No delimiters or instructions are provided to the agent to ignore potentially malicious content within the files it reads.
- Capability inventory: The skill frontmatter does not explicitly restrict tool usage; the instructions rely on the agent's ability to read local files.
- Sanitization: No sanitization or content validation steps are outlined for the data retrieved from the configuration files.
Audit Metadata