meta-meeting-to-workflow
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, prompt injections, or unauthorized data exfiltration attempts were detected. The skill logic is focused entirely on the stated purpose of meeting analysis.
- [DATA_EXPOSURE]: The skill ingests untrusted user content (
inputs.user_message). It mitigates risks by applyingxml_escapeandtruncatefilters before interpolating this data into prompts, preventing malformed content from breaking the workflow. - [COMMAND_EXECUTION]: The skill orchestrates tasks using internal capabilities (
summarize,github,sub-agent) rather than executing arbitrary shell commands or accessing sensitive local files. - [PROMPT_INJECTION]: The system instructions are well-defined and do not contain any bypass commands or attempts to override the agent's safety protocols.
- [EXTERNAL_DOWNLOADS]: The skill does not download external scripts, binaries, or unverifiable packages from the internet.
Audit Metadata