fireworks-training
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to facilitate model training using external datasets (JSONL), which represents an attack surface for indirect prompt injection.
- Ingestion points: User datasets are processed by validation scripts (e.g., in
references/choose-method.md) and training recipes. - Boundary markers: The skill provides validation logic to enforce structural constraints (role order, message presence) on ingested data.
- Capability inventory: The training environment includes capabilities for shell execution, network access (to Fireworks APIs), and model deployment.
- Sanitization: Scripts perform structural validation; however, content-level sanitization depends on platform-level guardrails and model-specific safety filters.
- [COMMAND_EXECUTION]: The skill provides numerous instructions for executing shell commands, including tool installation (
firectl), environment configuration (pip,venv), and execution of training recipes. These are standard developer operations for the intended use case. - [EXTERNAL_DOWNLOADS]: The skill references and downloads code from the official Fireworks AI cookbook repository on GitHub.
Audit Metadata