ea-connected-vault-setup

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process arbitrary user-controlled Markdown files from a vault to determine hierarchy and relationships.
  • Ingestion points: The agent reads the title, aliases, frontmatter, headings, outbound links, and body text of every note in the designated corpus (SKILL.md, Section 1).
  • Boundary markers: The instructions lack explicit boundary markers or instructions to ignore embedded commands within the processed notes, although it does define a specific scoring system for link creation.
  • Capability inventory: The agent has the capability to create new directories (Roots/), create new files (__Support/Bases/Children by Parents.base), and modify existing Markdown files by appending frontmatter and inline wikilinks (SKILL.md, Sections 2, 3, and 4).
  • Sanitization: No specific sanitization or filtering of the content within the notes is described; the agent is instructed to read enough body text to establish the subject, which could expose it to adversarial instructions placed inside those notes.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 12:38 PM
Security Audit — agent-trust-hub — ea-connected-vault-setup