skills/juxt/allium/propagate/Gen Agent Trust Hub

propagate

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted specifications and codebase data to automatically produce code, introducing a risk of indirect prompt injection.
  • Ingestion points: Processes external .allium files, project files from the codebase, and JSON output data from allium plan and allium model (SKILL.md).
  • Boundary markers: Absent. The instructions do not define strict boundary markers or instructions to isolate the parsed specifications from the system prompt execution flow.
  • Capability inventory: Possesses the capability to generate and write diverse test files, test fixtures, and scripts on the local file system for multiple ecosystems (TypeScript, Python, Rust, Go, Elixir).
  • Sanitization: Absent. There are no mechanisms specified to filter, sanitize, or safely escape the external text contents prior to test generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 01:17 PM
Security Audit — agent-trust-hub — propagate