Framework Detector

Pass

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: SAFE
Full Analysis
  • Indirect Prompt Injection (LOW): The skill identifies frameworks by scanning untrusted files within a project directory, which could allow an attacker to influence agent orchestration.
  • Ingestion points: Project configuration files (package.json, Gemfile, mix.exs, *.csproj) and source code files (.ts, .js, .rb, .ex, .cs, .razor) provided in the project path.
  • Boundary markers: No explicit delimiters or warnings for the agent to ignore instructions embedded within the scanned files are present.
  • Capability inventory: The detection output is intended to be consumed by SkillLoader, which dynamically loads additional tools or skills based on the classification result.
  • Sanitization: The skill uses weighted string and regex matching against a hardcoded list in framework-patterns.json, which restricts the output space but does not prevent an attacker from 'poisoning' a project to be misidentified as a specific framework.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 21, 2026, 10:22 AM