agentpmt-audit-logs
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is an official tool provided by AgentPMT for users to manage and audit their own account activity.
- [EXTERNAL_DOWNLOADS]: The documentation includes setup instructions that reference official AgentPMT repositories on GitHub (e.g., 'AgentPMT/agent-skills') for installing supplementary configuration skills.
- [DATA_EXFILTRATION]: The skill is designed to retrieve sensitive account data, including chat transcripts and tool execution history. This access is read-only, restricted to authorized groups, and is the primary declared function of the skill.
- [PROMPT_INJECTION]: By reading past chat transcripts and tool logs, the skill introduces a surface for indirect prompt injection. If a previous conversation contains instructions designed to influence an AI, those instructions would enter the agent's current context when logs are retrieved. This is a characteristic of audit tools and is mitigated by the skill's read-only nature and the platform's internal security controls.
Audit Metadata