post-outliner

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and act upon a 'topic brief' provided by the user, which constitutes untrusted input.
  • Ingestion points: Untrusted data enters the agent context through the topic brief mentioned in Phase 1 (ASSESS) and Phase 3 (GENERATE) of the instructions in SKILL.md.
  • Boundary markers: The skill uses specific markdown section headers (e.g., 'OUTLINE: [Working Title]') which provide minimal isolation from instructions that could be embedded within the user's brief.
  • Capability inventory: The skill's configuration in SKILL.md grants access to powerful system tools including Bash, Task, Write, and Edit.
  • Sanitization: There are no instructions or mechanisms provided to sanitize the user input or to direct the agent to ignore any embedded commands within the topic brief.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 05:55 AM
Security Audit — agent-trust-hub — post-outliner