map-product-system

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, such as repositories and document sets, to build system models. This creates a surface where malicious instructions embedded in the analyzed codebase could influence the agent's behavior. \n
  • Ingestion points: The skill reads user-provided repositories, documents, and specifications as evidence for building models (SKILL.md, README.md). \n
  • Boundary markers: The instructions mandate an evidence-labeling system (Known, Inferred, Proposed, Unknown) to distinguish source-derived facts from model-generated content, which provides context separation but does not fully block adversarial instructions. \n
  • Capability inventory: The skill uses file system read capabilities and provides architectural artifacts as output, often writing them to the project directory. \n
  • Sanitization: There are no explicit instructions for the agent to sanitize, escape, or filter external content before interpolation into its internal context.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 06:00 AM
Security Audit — agent-trust-hub — map-product-system