email-to-issue
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from Outlook email threads to create or update GitHub issues. Maliciously crafted email content could attempt to influence the agent's behavior during the conversion process.
- Ingestion points: The skill reads thread subjects and message bodies from Outlook via a Microsoft 365 connector.
- Capability inventory: The skill uses the
ghCLI to search issues, view repository visibility, and post comments. It also has the ability to reply, move, and flag emails via the connector. - Boundary markers: The instructions lack explicit delimiters or instructions to treat email content as data only, which could allow instructions within emails to be interpreted by the model.
- Sanitization: The skill mitigates risks through a mandatory manual approval step (Step 4) where the user reviews a summary table before any GitHub actions are taken. It also includes specific instructions to paraphrase confidential details when posting to public repositories.
Audit Metadata