skills/fw-ai-external/cookbook/debug/Gen Agent Trust Hub

debug

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to initialize the environment, specifically setting FIREWORKS_CLIENT_SOURCE and generating a unique FIREWORKS_SESSION_ID via python3 -c 'import uuid; print(uuid.uuid4())'. These are benign initialization steps.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to triage external data including job logs, metrics files (metrics.jsonl), and error messages. This creates a surface area for indirect prompt injection if those logs were to contain adversarial instructions. However, the skill explicitly instructs the agent to treat data as evidence for triage and includes safety-conscious rules to redact credentials and customer prose before any escalation.
  • [EXTERNAL_DOWNLOADS]: The instructions reference installing the fireworks-ai Python package. This is the official SDK for the vendor (fw-ai-external) and is considered a safe, primary tool for the skill's stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 06:47 AM