The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from security scan results and SBOMs, which are then used to inform sensitive actions such as creating or approving security exemptions.
Ingestion points: The skill uses harness_list and harness_get in SKILL.md to retrieve data for security_issue, scs_sbom, scs_artifact_component, and scs_compliance_result resources.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined to prevent the agent from following instructions potentially hidden in scan metadata.
Capability inventory: The skill utilizes harness_create and harness_execute tools in SKILL.md to manage security_exemption resources, allowing for state changes in the security posture.
Sanitization: There is no evidence of sanitization, validation, or filtering of the external security content before it is processed by the agent.

security-report