prefactor-skill-selector

Pass

Audited by Gen Agent Trust Hub on Feb 26, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of official developer tools and libraries, including @prefactor/cli and various SDK adapters, using standard package managers. These resources originate from the verified vendor organization prefactordev.
  • [COMMAND_EXECUTION]: Instructions are provided for running the prefactor CLI to manage agent environments and resources. These commands are part of the intended administrative functionality of the tool.
  • [PROMPT_INJECTION]: The instrumentation process creates a surface for capturing untrusted inputs and outputs from agent runs.
  • Ingestion points: Telemetry is collected at LLM and tool boundaries as specified in instrument-existing-agent-with-prefactor-sdk/SKILL.md.
  • Boundary markers: The documentation advises the use of package-specific span identifiers and execution contexts.
  • Capability inventory: The instrumentation logic transmits data to the Prefactor platform for observability.
  • Sanitization: The skill explicitly instructs users to implement redaction for sensitive information and enforce size limits on captured payloads to ensure secure data handling.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 26, 2026, 07:37 AM