summarize-meeting
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured templates and helper scripts for summarizing meeting notes. The included Python scripts perform standard data processing on local JSON files using standard libraries and contain no network operations, dynamic code execution, or credential access. \n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted meeting data (transcripts and notes), which is an inherent surface for indirect prompt injection. However, the skill follows best practices by providing rigid output templates and specific extraction rules that guide the agent towards safe processing. \n
- Ingestion points: The agent is instructed to process raw meeting notes, transcripts, and recordings as outlined in the SKILL.md Overview and When to Use sections. \n
- Boundary markers: The skill utilizes defined Markdown templates (assets/meeting_summary_template.md) to structure the output, which serves as a delimiter for extracted information. \n
- Capability inventory: The skill includes local Python scripts for data aggregation (scripts/action_item_tracker.py, scripts/decision_log_builder.py) and mentions integrations with external tools like Jira and Confluence. \n
- Sanitization: The methodology described in references/summarization-process.md emphasizes active extraction and synthesis, which acts as a manual filtering process for input data.
Audit Metadata