service-test
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses platform credentials (TFY_API_KEY) and the base URL from the environment or a local .env file. This data is used exclusively for authenticating requests to the TrueFoundry API and is not sent to unauthorized third parties.
- [EXTERNAL_DOWNLOADS]: References the 'truefoundry' Python package and official container images (e.g., from public.ecr.aws/truefoundrycloud). These are trusted vendor resources associated with the skill author.
- [COMMAND_EXECUTION]: Executes bash scripts (tfy-api.sh) and curl commands to interact with the TrueFoundry API and test service endpoints. These operations are restricted to the skill's primary purpose of service validation.
- [PROMPT_INJECTION]: The skill provides explicit instructions to treat all response bodies from tested services as untrusted third-party content. It specifically warns the agent not to follow or execute any instructions found in response bodies, which effectively mitigates indirect prompt injection risks.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Data is ingested via curl response bodies from service endpoints defined by the user.
- Boundary markers: The skill uses clear instructions to delimit how service response data should be handled.
- Capability inventory: The skill uses curl and the TrueFoundry CLI (tfy).
- Sanitization: The agent is instructed to only parse structured data (status codes and JSON fields) and ignore natural language instructions in response bodies.
Audit Metadata