fal-generate

Pass

Audited by Gen Agent Trust Hub on Jul 11, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill facilitates data transfer to fal.ai and fal.run domains for the purpose of AI content generation and file storage. This behavior is consistent with the skill's documented primary purpose.
  • [COMMAND_EXECUTION]: Local bash scripts use curl for network communication and jq for processing JSON data, which are standard utilities for API interaction.
  • [PROMPT_INJECTION]: The generate.sh script constructs JSON payloads by directly concatenating user-provided strings. This presents a potential surface for indirect prompt injection or JSON field injection. Ingestion points: Command line arguments for prompts and parameters. Boundary markers: None; input is interpolated directly. Capability inventory: Network requests via curl and file system access via stat. Sanitization: Absent; the script uses manual string concatenation without proper escaping.
  • [CREDENTIALS_UNSAFE]: The skill searches for the FAL_KEY environment variable and may also read it from .test_env files in parent directories. This is a common development pattern but requires users to ensure their environment files are appropriately protected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 11, 2026, 02:56 PM
Security Audit — agent-trust-hub — fal-generate