The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation describes the !exec YAML function, which enables the execution of arbitrary shell commands during configuration resolution.\n- [EXTERNAL_DOWNLOADS]: The skill supports fetching stack configurations from remote locations including GitHub, S3, GCS, OCI artifacts, and HTTPS URLs.\n- [REMOTE_CODE_EXECUTION]: Remote configuration files can contain the !exec function, potentially leading to execution of untrusted code if the remote source is compromised or malicious.\n- [DYNAMIC_EXECUTION]: Supports Go template processing (.yaml.tmpl) with Sprig functions, allowing for dynamic logic and environment variable access at runtime.\n- [INDIRECT_PROMPT_INJECTION]: The system processes external YAML data from remote sources that could influence agent behavior or execute commands.\n
Ingestion points: Remote URL imports (Git, S3, HTTPS) described in SKILL.md and references/import-patterns.md.\n
Boundary markers: Absent; configuration is deep-merged into the active state without explicit markers to ignore embedded instructions.\n
Capability inventory: !exec (shell execution), !terraform.output (state access), !store and !secret (secret resolution).\n
Sanitization: Absent; relies on standard YAML parsing and Go template rendering.

atmos-stacks