scrolls-update

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill facilitates project memory updates by providing cross-platform scripts (Bash and PowerShell) to identify changes in a git repository. The scripts use standard git commands (status, log, diff) to summarize session activity without performing any network operations or unauthorized data access.
  • [INDIRECT_PROMPT_INJECTION]: The skill relies on local configuration files, specifically STARTER.md, to determine how documentation should be updated. This creates an attack surface where instructions embedded in project files could influence agent behavior, though this is a fundamental part of its intended purpose to reflect project-specific conventions.
  • Ingestion points: Reads STARTER.md and other scroll files from the docs/.scrolls/ directory to guide update logic.
  • Boundary markers: Absent; the agent is instructed to treat STARTER.md as the authoritative source of truth for update rules.
  • Capability inventory: Execution of git commands and local file system write operations (overwriting and appending to markdown files).
  • Sanitization: None; the skill follows conventions defined in the project files as provided.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 05:49 PM
Security Audit — agent-trust-hub — scrolls-update