chunking-for-llms

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected in the skill's instructions or code examples.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external source code files for syntax-aware processing. This creates a surface where instructions embedded within source code could potentially influence an LLM. However, the skill only performs read-only chunking and does not autonomously execute the processed content.
  • Ingestion points: ts-pack process CLI and Python process() function in SKILL.md.
  • Boundary markers: Not explicitly defined in usage examples.
  • Capability inventory: Read-only source code processing and JSON output via ts-pack CLI and SDK.
  • Sanitization: Not explicitly mentioned in the documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 10:22 PM
Security Audit — agent-trust-hub — chunking-for-llms