Multica Runtime¶

Docker images that turn Claude Code, OpenCode and Codex into Multica agents.
One container, four environment variables, and you have an agent runtime that picks up tasks, works in a real repository, and keeps its session history across restarts.
cp .env.example .env # MULTICA_TOKEN, AI_BASE_URL, AI_API_KEY, AI_MODEL
docker compose up -d
That is the whole setup. No CLI installation, no Node/PHP/Flutter toolchain on your laptop, no per-agent configuration.
Why this exists¶
Running a Multica agent normally means installing an agent CLI, keeping it up to date, wiring it to a model endpoint, and repeating that on every machine. It gets worse when different tasks need different toolchains: a Laravel task needs PHP and three databases, a Flutter task needs the Android SDK, a QA task needs Playwright and k6.
These images put each stack in a container that:
- ships Claude Code, OpenCode and Codex, all pointed at your AI router;
- registers with Multica automatically and runs the daemon in the foreground;
- runs as a non-root user (uid 1000) on Debian 13;
- keeps sessions on a volume, so a restart does not lose context;
- is built in CI for
linux/amd64andlinux/arm64, with every tool pinned.
What is included¶
| Image tag | What you get |
|---|---|
:base |
Node 26 LTS, Python 3, git/gh, Chromium, build tools, Claude Code, OpenCode, Codex CLI, Multica CLI |
:node |
pnpm, yarn, TypeScript, Playwright, vitest, eslint, prettier |
:python |
uv, Poetry, ruff, mypy, pytest, C headers |
:php |
PHP 8.4, Composer, Laravel, MariaDB + PostgreSQL + Redis inside the container |
:go |
Go toolchain, golangci-lint |
:flutter |
Flutter, Android SDK, JDK 21, Gradle |
:devops |
Terraform, AWS CLI, kubectl, Helm, Docker CLI, Wrangler, yq |
:qa |
Playwright + MCP, axe-core, newman, k6, pytest |
:design |
ImageMagick, Playwright, Penpot MCP + plugin |
All tags come from one package: ghcr.io/proreka/multica-runtime.
The configuration surface¶
Only these four are required:
MULTICA_TOKEN=... # https://multica.ai/settings?tab=tokens
AI_BASE_URL=https://... # any Anthropic-compatible endpoint
AI_API_KEY=...
AI_MODEL=...
Both agent CLIs are configured from those values. Native variables
(ANTHROPIC_*) still work and win when both are set.
→ Full configuration reference
Documentation¶
- Quickstart — from zero to a registered runtime
- Configuration — every variable, model tiers, routers
- Variants — which image for which job
- Architecture — what happens at container start
- Autoscaling — scale roles between 0 and 1 from pending tasks
- Troubleshooting — symptoms and fixes
- Security — what the container trusts
Requirements¶
- Docker 24+ (Compose v2) or any orchestrator that can run a container
- A Multica token
- An Anthropic-compatible endpoint. OpenCode can also use an OpenAI-compatible
router (
/v1/modelsis probed withAI_PROTOCOL=auto); Claude Code still needs an Anthropic-compatible surface.
No GPU, no privileged mode, no Docker socket — except for the optional Swarm
autoscaler, which needs it to call docker service scale.
Status¶
Claude Code, OpenCode and Codex CLI are supported now. Qwen Code, Grok Build, DeepSeek Harness, Pi, Oh-My-Pi, Hermes Agent, OpenClaw and more are on the roadmap.
License¶
Apache-2.0. See LICENSE.