doc-coauthoring

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection vulnerability surface detected due to the processing of untrusted external content into agent-controlled artifacts.
  • Ingestion points: SKILL.md (Stage 1: Context Gathering) instructs the agent to pull data from 'team channels', 'shared documents', and user-provided 'info dumps'.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to isolate untrusted content from the agent's core logic.
  • Capability inventory: The skill utilizes create_file and str_replace tools to modify the local filesystem based on the ingested content and invokes sub-agents in Stage 3 (Reader Testing) with the generated text.
  • Sanitization: There is no requirement or logic for sanitizing or escaping the external content before it is used to draft documents or passed to other LLM instances.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 09:47 PM
Security Audit — agent-trust-hub — doc-coauthoring