session-history
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_&_EXFILTRATION]: The skill uses a tool called
memory_sessionsto retrieve session history metadata (IDs, timestamps, status, and observation counts). This is a legitimate use of project metadata for project management tasks. - [CREDENTIALS_UNSAFE]: The
TROUBLESHOOTING.mdfile correctly instructs users and developers to manage secrets using.envfiles and warns against displaying them or sending local secrets to remote URLs. It specifically mentions~/.agentmemory/.env, which is a standard and safe practice for local credential management. - [REMOTE_CODE_EXECUTION]: No remote code execution or external package installations were detected. The skill uses pre-existing tools and optional REST API fallbacks to local endpoints (
localhost,127.0.0.1). - [INDIRECT_PROMPT_INJECTION]: While the skill processes data from past sessions (which could theoretically contain untrusted content), it focuses on metadata (titles, decision types, and observation counts) rather than executing or interpreting the content of those sessions. This limits the attack surface.
- [PROMPT_INJECTION]: The instructions include strong 'Anti-patterns' and 'Checklist' sections that reinforce the agent's adherence to tool outputs, preventing the agent from inventing or hallucinating data.
Audit Metadata