The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies ingestion points for data from external sources that are used to drive configuration and verification, creating a surface for potential indirect prompt injection attacks.\n
Ingestion points: The skill reads cluster state (pod IPs, ClusterIPs) and Cloud API data (load balancer firewall rules, container registry digests) as described in references/acceptance.md and references/failure-catalogue.md.\n
Boundary markers: No specific delimiters or instructions to ignore embedded commands are documented for the data received from these external APIs.\n
Capability inventory: The skill utilizes significant system capabilities including kubectl for manifest application and curl for interacting with the NetBird management API as described in SKILL.md.\n
Sanitization: While the skill performs validation on CIDR membership, it does not explicitly describe sanitization or escaping of data retrieved from cloud and cluster APIs before it is used in automation logic.\n- [EXTERNAL_DOWNLOADS]: The skill relies on external software components and images from known and vendor-specific sources.\n
Evidence: References verified container images for netbird-server, reverse-proxy, traefik, and kaniko from their respective official repositories in references/pins.md.\n
Evidence: Mentions a reference implementation and package skill hosted in the author's (getcolors) GitHub repository.

agent-network-doks