doc-coauthoring

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill is designed to ingest external user content, shared documents, and team channel data, which presents a potential surface for indirect prompt injection if the ingested text contains instructions meant to override agent behavior.\n
  • Ingestion points: User text dumps, shared files, templates, and connected channels/threads specified in SKILL.md.\n
  • Boundary markers: The workflow does not explicitly implement specialized boundary markers or delimiters around untrusted external content.\n
  • Capability inventory: The skill utilizes file creation and replacement tools (create_file, str_replace) and supports sub-agent invocation for document validation.\n
  • Sanitization: No explicit sanitization or instruction filtering is applied to the ingested content before processing or writing to artifacts.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 11:57 PM
Security Audit — agent-trust-hub — doc-coauthoring