yzr-memory-management
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill includes a proactive security utility,
tools/memory_lint.py, which scans the memory files for hardcoded credentials, secrets, and sensitive tokens (using patterns for AWS, GitHub, etc.) and warns the user if any are found. - [SAFE]: All automated workflows (capture, audit, init) follow a 'propose-then-execute' model, ensuring that the agent does not modify or delete memory files without explicit user confirmation, which prevents unauthorized autonomous changes.
- [SAFE]: Python scripts used for management tasks (
memory_init.py,memory_lint.py) rely exclusively on the Python standard library and do not perform any external network requests or execute untrusted remote code. - [INDIRECT_PROMPT_INJECTION]: The skill manages a persistent memory system where project rules are stored in markdown files. While this creates a potential surface for indirect prompt injection, it is the primary intended function of the skill and the risk is significantly mitigated by the requirement for human review of all proposed memory updates.
- Ingestion points: Markdown files located in the
MEMORY/directory. - Boundary markers: None explicitly mentioned for the content of the memory files themselves.
- Capability inventory: The skill can read and write files within the project directory using its management scripts.
- Sanitization: Structural linting and secret scanning are performed, but content is not filtered for natural language injection patterns.
Audit Metadata