Customization And Images
Agirunner lets operators move beyond one default task image.
That matters because real execution environments often need more than a generic shell. Teams may need language runtimes, package managers, system libraries, certificates, or security posture that match their actual work.
What This Covers
Section titled “What This Covers”- the runtime image that hosts the long-running execution service
- task images used for specialist execution
- customization and build flows exposed by the platform
- the relationship between published images and execution-environment records
Why It Matters
Section titled “Why It Matters”The goal is not just customization for its own sake. The goal is to make execution environments explicit, reviewable, and reproducible.
Two Image Layers
Section titled “Two Image Layers”Agirunner intentionally separates the image that hosts the agent loop from the image that hosts specialist task execution.
- The long-running runtime image is a lightweight Alpine-based image that carries the specialist or orchestrator agent loop, platform connectivity, and execution machinery.
- The specialist execution environment is where heavier language runtimes, package managers, and task-specific system dependencies belong.
That split keeps the control loop lean while still giving teams room to run real work in richer execution environments.
Built-In Starter Catalog
Section titled “Built-In Starter Catalog”The platform seeds a built-in starter catalog of execution environments. Today that includes Debian, Ubuntu, Alpine, Fedora, Python, Node, Go, Rust, Java, and PHP starters, with Debian Base seeded as the default environment for a fresh tenant.
Operators manage those records from Environments, attach them to specialist roles from Specialists, Skills, And Models, and can replace them with custom images as needed.