refiners-update

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill follows a legitimate workflow for document maintenance and orchestration without any detected malicious patterns.
  • [COMMAND_EXECUTION]: The skill coordinates revisions by calling other specialized refiner skills (such as /architecture-refiner and /language-idioms-refiner), which is the standard mechanism for complex task delegation in agentic workflows.
  • [DATA_EXPOSURE]: The skill reads project configuration (.lattice/config.yaml) and existing markdown standards in the .lattice/standards/ directory to determine the current project state. These read operations are scoped to the local project environment.
  • [PROMPT_INJECTION]: The skill processes user-provided descriptions of changes and interpolates them into document footers as audit notes. While this ingestion of untrusted data represents an indirect prompt injection surface, it is a primary intended functionality of the tool. Mandatory Evidence: 1. Ingestion points: User-provided change reasons in Step 2. 2. Boundary markers: Absent for the interpolated reason string. 3. Capability inventory: Local file read/write and skill invocation. 4. Sanitization: No explicit filtering or validation of user-provided strings before inclusion in documents.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 03:44 PM
Security Audit — agent-trust-hub — refiners-update