incident-init
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to process untrusted data from external sources such as incident alerts and monitoring bot outputs.
- Ingestion points: Untrusted data enters the context through channel names, alert text (e.g., from PagerDuty or Datadog), and recent channel message history as described in
SKILL.md. - Boundary markers: The instructions include a specific security directive to use external data only to identify teams and "never as instructions," which is a proactive measure against prompt injection.
- Capability inventory: The skill possesses capabilities to post briefings, pin messages, write to channel memory, and generate charts using the
datavizskill. - Sanitization: The instructions explicitly prohibit the copying of sensitive data such as tokens or personal information into the channel memory, mitigating risk in the event of an injection.
Audit Metadata