stream-chain

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODECOMMAND_EXECUTION
Full Analysis
  • [NO_CODE]: The skill consists exclusively of markdown documentation. No implementation code (Python, JavaScript, etc.) was provided for review.- [COMMAND_EXECUTION]: The skill documentation details the use of a command-line interface (claude-flow) to perform complex tasks, including automated code refactoring and performance optimization.- [PROMPT_INJECTION]: The skill describes a multi-agent pipeline that processes external inputs such as API responses and source code files, which constitutes a surface for indirect prompt injection.
  • Ingestion points: External API data and local source code files in the src/ directory.
  • Boundary markers: Documentation does not define specific delimiters or instructions to ignore embedded commands within the processed data.
  • Capability inventory: Filesystem modification through refactoring actions, network access via data enrichment steps, and general CLI command execution via the claude-flow utility.
  • Sanitization: No input validation or output sanitization mechanisms are described in the documentation or provided examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 05:44 AM