orbitant-learning-report

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates on local data within a specific subdirectory of the agent's workspace to generate learning summaries. It follows best practices for data handling within its scope.
  • [PROMPT_INJECTION]: The skill processes learning logs which represent a potential surface for indirect prompt injection. The impact is minimal because the skill is strictly local and lacks capabilities to perform sensitive actions based on the content of these logs.
  • Ingestion points: Files in ~/.claude/learning/logs/ and ~/.claude/learning/lessons/.
  • Boundary markers: No delimiters or safety warnings are implemented for processing these logs.
  • Capability inventory: Basic file read and write operations within the ~/.claude/learning/ directory.
  • Sanitization: None.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 11:49 PM
Security Audit — agent-trust-hub — orbitant-learning-report