The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the valet CLI and git to manage services, configuration, and deployments. These commands are integral to the skill's primary function of infrastructure orchestration.
[REMOTE_CODE_EXECUTION]: The skill includes instructions to install the valet CLI using shell-piped installers (curl | sh and irm | iex) from the vendor's domain (valet.dev). As this targets the official vendor infrastructure for the required tooling, it is documented as a standard setup procedure.
[INDIRECT_PROMPT_INJECTION]: The skill defines a framework for creating agents whose behavior is governed by external files such as SOUL.md and organization-level skills like governance. This architecture presents a surface for indirect prompt injection if these configuration files are sourced from untrusted repositories.
Ingestion points: Behavioral instructions are ingested from SOUL.md files during agent creation and from the governance skill fetched from the organization's Valet account.
Boundary markers: The skill encourages structured sections (Purpose, Workflow, Guardrails) and explicit scope rules for webhook handlers to maintain operational boundaries.
Capability inventory: The agent possesses the capability to execute shell commands, deploy web applications, and set organization-wide environment variables.
Sanitization: The skill recommends using valet manifests validate to verify the structure and content of configuration files before deployment.

valet