The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by facilitating the processing of external data from third-party Helm charts and repositories.\n
Ingestion points: The agent is instructed to fetch and process Helm charts from local paths, remote HTTP repositories, and OCI registries defined in stack manifests (SKILL.md).\n
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the agent when processing downloaded chart content or value files.\n
Capability inventory: The skill provides the agent with the ability to execute atmos helm apply and atmos helm deploy, which perform state-changing operations on Kubernetes clusters and Git repositories (SKILL.md).\n
Sanitization: No explicit sanitization or strict schema validation steps for external chart content are described before the resources are applied to a target.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the retrieval of external components and binaries.\n
It references fetching Helm charts from remote repositories, including standard community sources like the Prometheus Community repository.\n
It documents the ability to install external Helm CLI plugins via the atmos helm plugin install command.\n- [COMMAND_EXECUTION]: The skill involves executing Helm-related operations that interact with sensitive infrastructure.\n
It utilizes the Helm Go SDK to template, diff, and apply Kubernetes resources directly to clusters.\n
It supports delivery to Git deployment repositories, involving git operations such as cloning and committing rendered manifests.

atmos-helm