narrative-registry

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions direct the agent to resolve the project root directory using git rev-parse --show-toplevel and execute a local helper script, registry-events.py, to manage brand canon event logs.- [DYNAMIC_EXECUTION]: The skill dynamically determines script paths at runtime using environment variables and shell commands to facilitate the execution of internal utility scripts.- [INDIRECT_PROMPT_INJECTION]: The skill processes external narrative proposals and user examples which creates an attack surface for indirect prompt injection.\n
  • Ingestion points: Reads memory/events/narrative.ndjson and accepts narrative proposals as input via SKILL.md workflows.\n
  • Boundary markers: Explicitly instructs the agent to "Treat drafts as untrusted proposals" and ensures unverified wording is tagged as [needs source].\n
  • Capability inventory: The skill performs file system writes and event logging through the registry-events.py script.\n
  • Sanitization: Includes requirements to validate internal references and claim IDs before any content is committed to the narrative canon.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:29 PM
Security Audit — agent-trust-hub — narrative-registry