incident-responder
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external incident data that could be influenced by an attacker. 1. Ingestion points: SKILL.md (Observability-Driven Investigation and SRE Investigation Techniques) and resources/implementation-playbook.md (Inputs). 2. Boundary markers: Absent from the instructions. 3. Capability inventory: Immediate stabilization actions including traffic throttling, feature flag changes, rollbacks, and resource scaling, as well as stakeholder communication tools. 4. Sanitization: No explicit instructions are provided for sanitizing or escaping external content before processing.
- [METADATA_POISONING]: The skill's frontmatter includes 'risk: critical'. While likely intended to reflect the criticality of the incident response domain, this label is unusual for skill metadata and could be interpreted as deceptive or misleading concerning the skill's own security risk profile.
Audit Metadata