Standard Runtime Registry
The complete language-version-to-image mapping table for AEGIS StandardRuntime — supported versions, base images, pre-installed toolchains, and the version support lifecycle.
Standard Runtime Registry
When an agent manifest specifies spec.runtime.language and spec.runtime.version, the AEGIS orchestrator resolves them to a deterministic, pinned Docker image at execution time. The canonical mapping is defined in runtime-registry.yaml, which is committed to the orchestrator repository and loaded at daemon startup.
Using StandardRuntime means you never specify a Docker image directly — AEGIS selects a vetted, pinned image on your behalf.
Runtime Matrix
| Language | Version | Docker Image | Base OS | Package Manager | Status |
|---|---|---|---|---|---|
python | 3.11 | python:3.11-slim | Debian 12 | pip | Supported |
python | 3.10 | python:3.10-slim | Debian 12 | pip | Supported |
python | 3.9 | python:3.9-slim | Debian 12 | pip | Deprecated — Phase 2 removal planned; migrate to 3.10 or 3.11 |
javascript | 20 | node:20-alpine | Alpine 3.18 | npm | Supported (Current LTS) |
javascript | 18 | node:18-alpine | Alpine 3.18 | npm | Supported (Legacy LTS) |
typescript | 5.1 | node:20-alpine | Alpine 3.18 | npm, ts-node | Supported |
go | 1.21 | golang:1.21-alpine | Alpine 3.18 | go | Supported |
go | 1.20 | golang:1.20-alpine | Alpine 3.18 | go | Supported (Legacy) |
rust | 1.75 | rust:1.75-alpine | Alpine 3.18 | cargo | Supported |
rust | 1.74 | rust:1.74-alpine | Alpine 3.18 | cargo | Supported (Legacy) |
Manifest Usage
spec:
runtime:
language: "python" # Required (StandardRuntime)
version: "3.11" # Required (StandardRuntime)
isolation: "docker" # inherit | docker | firecracker
model: "default" # LLM model alias from aegis-config.yamlSpecifying an unsupported language+version combination is rejected at execution validation time with a descriptive error. There is no fallback to a default version.
Pre-Installed Toolchains
Each StandardRuntime image includes the language's standard package toolchain. You can install additional dependencies inside your agent's bootstrap or using cmd.run tools.
Python (python:3.x-slim, Debian 12)
pip(Python package manager)apt(Debian package manager, for OS-level packages)
# Install a Python package inside the agent
pip install requests numpy
# Install an OS package
apt-get install -y libssl-devTip: If you need more than a few OS packages, consider a Custom Runtime to avoid repeated installs at execution time.
JavaScript / TypeScript (node:x-alpine, Alpine 3.18)
npm(Node.js package manager)apk(Alpine package manager)
For TypeScript (5.1), the orchestrator sets the TYPESCRIPT_VERSION=5.1 environment variable. Your bootstrap or Dockerfile should install the TypeScript toolchain:
npm install -g typescript ts-nodeGo (golang:1.x-alpine, Alpine 3.18)
gobinary and build toolchainapk(Alpine package manager)
# Install a Go dependency
go get github.com/some/libraryRust (rust:1.x-alpine, Alpine 3.18)
cargo(Rust package manager and build tool)rustc(Rust compiler)- Rust standard library
# Add a crate dependency
cargo add serdeBootstrap Mechanism
For all StandardRuntime agents, the orchestrator:
- Resolves
language+version→ Docker image via the registry - Creates the container:
docker run -d --name aegis-<execution_id> <image> tail -f /dev/null - Checks whether
/usr/local/bin/aegis-bootstrapalready exists inside the container - If absent, uploads the orchestrator's
assets/bootstrap.pyto/usr/local/bin/aegis-bootstrap - Executes
python /usr/local/bin/aegis-bootstrap <prompt>
The bootstrap script connects to AEGIS_ORCHESTRATOR_URL → /v1/dispatch-gateway and runs the 100monkeys iteration loop (generate → execute → validate → refine).
Daemon Configuration
The orchestrator loads the registry from disk at startup via spec.runtime.runtime_registry_path in aegis-config.yaml. The default value is "runtime-registry.yaml" (relative to the daemon's working directory), which matches the file committed to the orchestrator repository root — no configuration is needed for the common case.
# aegis-config.yaml
spec:
runtime:
# Path to the StandardRuntime registry YAM.
# Default: "runtime-registry.yaml" (relative to daemon working directory)
runtime_registry_path: "runtime-registry.yaml"Startup behaviour: If the file is missing or contains a parse error, the daemon hard-fails at boot with a descriptive error message. It does not fall back silently, because a missing registry makes every StandardRuntime agent execution impossible.
Custom path: Set runtime_registry_path to an absolute path (e.g., "/etc/aegis/runtime-registry.yaml") when running the daemon from a directory other than the repository root, or when deploying via container images where the working directory differs from the source tree.
Version Support Lifecycle
Each language version in the registry follows a support lifecycle:
| Status | Meaning |
|---|---|
| Supported | Maintained; receives security patches from the upstream language project |
| Deprecated | Still functional, but officially end-of-life upstream; logs a warning at execution time |
| Removed | No longer accepted in manifests; execution validation fails |
Version removals are announced two phases in advance: a deprecation warning phase (still functional) is followed by a removal phase. Agents using deprecated versions will see warnings in execution logs; migrate proactively.
Currently deprecated: python: "3.9" — official upstream EOL. Manifests using this version will execute but log deprecation warnings. Migrate to python: "3.10" or python: "3.11".
Choosing Between StandardRuntime and CustomRuntime
| Need | Use |
|---|---|
| Supported language and version | StandardRuntime — no image to build or maintain |
OS-level packages (e.g., ffmpeg, graphviz) | CustomRuntime |
| Language version not in the registry | CustomRuntime |
| Pre-compiled binaries in the image | CustomRuntime |
| Proprietary or private dependencies | CustomRuntime |
See Also
- Custom Runtime Agents — Using your own Docker image
- Container Registry & Image Management — ImagePullPolicy, private registry authentication, and image caching
- Agent Manifest Reference —
spec.runtimefield definitions
Agent Manifest Reference
Complete specification for the AgentManifest YAML format (v1.0) — schema, field definitions, examples, and validation configuration.
Workflow Manifest Reference
Complete specification for the WorkflowManifest YAML format — schema, field definitions, all six state kinds including ContainerRun for CI/CD steps, consensus strategies, and complete examples.