debug
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to manage environment variables (
FIREWORKS_SESSION_ID,FIREWORKS_CLIENT_SOURCE) and utilizes thefirectlCLI tool to query job statuses. These operations are restricted to triage and diagnostic purposes within the vendor's ecosystem. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as job logs, error messages, and status outputs from the Fireworks API (referenced in
references/triage-paths.md). This constitutes an ingestion surface where external content could theoretically influence the agent's behavior. However, the skill explicitly mandates a read-only approach and requires the redaction of customer data and credentials before any escalation, mitigating potential risks. - [SAFE]: The skill includes privacy safeguards, such as a telemetry notice and specific rules against sending raw errors or customer prose as journey telemetry. These design choices prioritize data protection during the debugging process.
Audit Metadata