The Agent Skills Directory

[SAFE]: The skill provides standard documentation and operational procedures for Flux CD. It encourages security best practices such as pinning Helm chart versions and image digests, and specifically instructs the agent to avoid hardcoded credentials by using Kubernetes secret references.
[SAFE]: All external references point to official project documentation and repositories for Flux CD (fluxcd.io and github.com/fluxcd). These are recognized well-known sources for the technology described.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by instructing the agent to ingest and analyze output from the cluster environment.
Ingestion points: The agent is directed to read flux logs, kubectl get events, and status tables from flux get (SKILL.md, gemini.md, gpt.md).
Boundary markers: None present in the instructions.
Capability inventory: The skill utilizes shell execution for flux and kubectl CLI tools.
Sanitization: No specific sanitization or escaping of external content is defined.

fluxcd