scene-sequencing

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements narrative analysis using a local Deno script (analyze-scene.ts) that only requests read permissions (--allow-read). It does not perform network operations, access sensitive credentials, or attempt to escalate privileges.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted fiction drafts provided by users, which is inherent to its purpose. While this creates a surface for indirect prompt injection, the primary analysis tool uses deterministic regular expressions which are not susceptible to instruction override.
  • Ingestion points: Scene text processed via scripts/analyze-scene.ts and the agent's scene review instructions.
  • Boundary markers: No specific delimiters or safety warnings for embedded instructions are present in the text ingestion process.
  • Capability inventory: The skill utilizes local file read access for analysis and writes session data to local configuration files.
  • Sanitization: User-provided text is processed for pattern matching without explicit escaping or sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 11:09 AM
Security Audit — agent-trust-hub — scene-sequencing