Shiro Automation v2026.06.06.10

Welcome to Shiro Docs

Shiro runs inside your existing CI runner as a Docker image — no new infrastructure, no agents. Define workflows in JSON, trigger them from GitLab CI or GitHub Actions, and get AI-native capabilities out of the box.

CI users — start hereGitLab CI · GitHub Actions

Use the Shiro Docker image directly in your pipeline. No install step needed.

# .gitlab-ci.yml
shiro:
  image: ghcr.io/rajitk13/shiro-automation:latest
  script:
    - shiro run -workflow .shiro/workflows/my-workflow.json -state-store gitlab
Full setup guide (3 steps)

What is Shiro?

Shiro is a production-ready Go-based workflow runtime designed to run inside existing CI runners (GitLab, Jenkins, GitHub Actions, Kubernetes Jobs). It executes ephemeral workflows with DAG-based scheduling, supports reusable modules and integrations, and enables AI-assisted workflows with multiple provider support including Ollama, OpenAI, and custom endpoints.

Key Features

  • Simplified CLI — Intuitive commands with auto-detection (shiro init, shiro run, shiro add module)
  • Portable Runtime — Single binary that runs in any CI environment
  • DAG Execution — Topological sorting with dependency management
  • Module System — Pluggable architecture with GitHub marketplace integration
  • Auto-Discovery — Search and install modules from official GitHub repository
  • AI Providers — Support for Ollama, OpenAI, and custom endpoints
  • State Storage — Modular backends (GitLab artifacts, filesystem, memory)
  • Human-in-Loop — GitLab-native manual approval workflows with Slack notifications

Add to your CI today

Three files. One push. Shiro runs in your existing GitLab CI or GitHub Actions pipeline — no new infrastructure required.

Architecture Overview

Trigger Adapters (GitLab/Jenkins/GitHub)
              ↓
      Workflow Runtime (Go)
              ↓
         DAG Executor
              ↓
      Module/Plugin System
              ↓
    AI / Integrations / Compute