The Agent Skills Directory

Session Transcript Processing: The skill reads and analyzes agent transcripts stored on the local filesystem. These logs naturally contain user prompts, code snippets, and tool outputs from the analyzed session, which may include sensitive data or credentials if they were present in the original interaction. This access is necessary for the skill's diagnostic purpose.
Privacy Safeguards: The skill emphasizes transparency by requiring the agent to display a privacy notice to the user. It explicitly instructs the user to review and redact the generated report before sharing it, mitigating the risk of accidental data exposure.
Indirect Prompt Injection Surface: Since the skill ingests and summarizes logs from previous agent sessions, it represents an indirect prompt injection surface. If a session transcript contains adversarial instructions, the agent summarizing the report could potentially be influenced by them. This is a common security consideration for any tool that processes and interprets external data or logs.

winui-session-report