snapshot-writer

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill performs its stated function of literature summarization using local workspace artifacts.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from outline/outline.yml and papers/core_set.csv to generate snapshots. \n
  • Ingestion points: Data enters via outline/outline.yml and papers/core_set.csv in SKILL.md and scripts/run.py.\n
  • Boundary markers: The prompt uses role-based separation (Snapshot Editor, Pointer Curator) but lacks explicit delimiters for untrusted input data.\n
  • Capability inventory: The skill has file read/write access to the local workspace and does not utilize network or shell execution tools.\n
  • Sanitization: The instructions explicitly forbid the creation of fake citations and limit data sources to specific local files, mitigating the risk of data-driven hallucination.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 06:47 AM