agents-memory
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill includes several utility scripts (audit_repo.sh, audit_portfolio.sh, lint_claude_memory.sh, compare_blocks.sh) designed to perform static analysis of project documentation. These scripts operate locally, are read-only regarding the codebase being analyzed, and use standard Unix tools to identify stale paths and configuration errors.
- [CREDENTIALS_UNSAFE]: The lint_claude_memory.sh script contains a security feature that uses regular expressions to scan for hardcoded credentials (such as OpenAI/Anthropic API keys and private keys) within memory files, alerting the user if they are found. This is a security-enhancing feature.
- [EXTERNAL_DOWNLOADS]: The documentation and data/sources.json file reference a wide array of official documentation (Anthropic, OpenAI, GitHub) and reputable community resources. These links are for informational purposes and do not trigger unauthorized remote code execution.
- [COMMAND_EXECUTION]: The skill documentation provides examples of standard development commands (e.g., npm test, git status) and suggests using them for verification. The included scripts execute local analysis commands that are part of the skill's core functionality.
- [PROMPT_INJECTION]: The instructions utilize standard adherence-strengthening markers like IMPORTANT: to guide the agent's behavior. No patterns were found that attempt to bypass system safety filters or override global agent constraints.
Audit Metadata