configure
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFE
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill follows security best practices for secret management. It explicitly instructs the agent to check for the presence of the
FIREWORKS_API_KEYwithout printing it and prohibits asking for keys in chat or reading them from unsafe credential files. It recommends secure terminal entry or gitignored.envfiles. - [EXTERNAL_DOWNLOADS]: The skill references and fetches data from official Fireworks AI domains (
fireworks.aianddocs.fireworks.ai). These are well-known, vendor-controlled services used for documentation and model catalog retrieval, presenting no security risk. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied datasets (JSONL format) for model training. While this presents a potential attack surface, the instructions mandate local validation of data (schema, roles, and content) before any upload or processing occurs, mitigating the risk of processing malicious instructions embedded in data.
- [COMMAND_EXECUTION]: The skill uses the
firectlCLI and Fireworks Python SDK for administrative tasks. It includes a safety guardrail wherefirectlis configured to block mutating commands by the agent, ensuring that any modifications require the agent to use the more controlled SDK or REST API paths. - [TIME_DELAYED_CONDITIONAL]: The skill contains logic to check a specific date and time (
2026-12-01T00:00:00-08:00) to determine the eligibility for a promotional pricing offer ($0 SFT). This is a benign business rule and does not constitute a malicious trigger. - [DYNAMIC_EXECUTION]: The skill uses simple Python one-liners for utility tasks, such as generating a UUID for session tracking (
python3 -c 'import uuid; print(uuid.uuid4())'). This is a standard and safe use of dynamic execution for runtime metadata generation.
Audit Metadata