ac-memory-manager

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill implements a persistent memory system that stores arbitrary data, creating a surface for indirect prompt injection if the agent subsequently interprets retrieved memories as instructions.
  • Ingestion points: scripts/memory_manager.py (via the store and retrieve methods).
  • Boundary markers: Absent. Data is stored and retrieved as raw values without delimiters or instructions to ignore embedded commands.
  • Capability inventory: Local file system write access (_save_memory) to the .claude/memory/ directory.
  • Sanitization: Absent. The skill serializes data directly to JSON without filtering for potential injection patterns.
  • [Data Exposure & Exfiltration] (SAFE): The script manages data locally within the project directory (.claude/memory). No network calls (e.g., requests, curl) or access to sensitive system-level file paths (e.g., ~/.ssh) were detected.
  • [Unverifiable Dependencies] (SAFE): The implementation relies exclusively on the Python standard library. No external package managers or remote code execution patterns are present.
  • [Command Execution] (SAFE): There are no uses of os.system, subprocess, or similar utilities that execute shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:28 PM