knowledge-ops

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The ingestion workflow involves processing data from untrusted external sources, creating a surface for indirect prompt injection where malicious instructions in the data could influence agent behavior.
  • Ingestion points: SKILL.md describes ingesting documents, conversations, browser bookmarks, and GitHub activity events.
  • Boundary markers: Absent; no specific delimiters or ignore instructions are defined for processing external content.
  • Capability inventory: SKILL.md references the ability to write to local memory files, commit and push to Git repositories, and use MCP tools for memory management.
  • Sanitization: Instructions mention redacting secrets but do not specify sanitization or escaping of natural language content to prevent instruction injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 04:00 AM
Security Audit — agent-trust-hub — knowledge-ops