absorb

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from git repositories and local paths to extract AI-harness practices, creating a surface for embedded instructions to influence the agent.
  • Ingestion points: External file content is read and mapped in SKILL.md (Workflow steps 1 and 2).
  • Boundary markers: The skill includes 'Hard Rules' to ignore README claims, but it lacks specific prompt delimiters to separate external data from agent instructions.
  • Capability inventory: The skill has capabilities to clone repositories, read arbitrary files, and write analysis reports to the local filesystem (.ai/absorb/).
  • Sanitization: There is no evidence of sanitization or escaping of the content extracted from external targets before it is processed by the agent.
  • [EXTERNAL_DOWNLOADS]: The skill downloads content from external git repositories provided by the user.
  • Evidence: SKILL.md workflow step 1: 'If it is a git URL, clone or update it only with user awareness'.
  • Context: This behavior is the primary intended purpose of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 08:12 PM
Security Audit — agent-trust-hub — absorb