story-consistency-monitor
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill is designed to ingest and process untrusted external data (user-written story chapters and character profiles) which could contain malicious instructions.
- Ingestion points: Processes content from
characters/,worldbuilding/,spec/tracking/timeline.json, and current chapter text. - Boundary markers: No explicit delimiters or instructions are defined to separate narrative text from agent instructions.
- Capability inventory: Uses
ReadandGreptools. While the documentation mentions 'auto-fixes', the permitted toolset does not explicitly include file-writing permissions, limiting impact to suggested output. - Sanitization: No evidence of sanitization or validation for the content being monitored.
- [Command Execution] (SAFE): The mentioned integrations (
/write,/analyze,/track) appear to be part of a structured internal environment rather than shell-level command execution. No unauthorized subprocess calls were detected. - [Data Exposure] (SAFE): File access is scoped to specific project-related directories. There is no evidence of attempts to access sensitive system files (e.g., SSH keys, environment variables) or network exfiltration.
Audit Metadata