on-call-handoff-patterns
Pass
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes shell command templates for system administration tasks.
- Evidence: Examples using
kubectl,psql, andredis-cliare provided in the 'Quick Reference' section ofSKILL.md. - Evidence: Includes a destructive command
redis-cli FLUSHDBfor emergency cache clearing. - [EXTERNAL_DOWNLOADS]: The skill references external resources and links to standard SRE documentation.
- Evidence: Links to the Google SRE book, PagerDuty, and generic infrastructure placeholders (Grafana, Wiki) are included for educational and operational context.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its handling of engineer-written handoff summaries.
- Ingestion points: The agent is instructed to process shift handoff documents, incident reports, and investigation logs, which are provided via the templates in
SKILL.md. - Boundary markers: The templates do not implement explicit delimiters or warnings to ignore embedded instructions in the 'Context' or 'Next Steps' sections.
- Capability inventory: The skill grants the agent capabilities to interact with Kubernetes clusters, databases, and Redis caches using powerful administrative commands.
- Sanitization: There is no evidence of sanitization or validation of the data processed from handoff documents before it influences the agent's actions.
Audit Metadata