Documentation

Callahan CI

Callahan CI is an open-source, AI-native, serverless CI/CD platform that runs entirely on your machine. No cloud account. No Kubernetes cluster. No plugin graveyard. One command and you're building.

Installation

Callahan runs on Linux, macOS, and Windows (WSL2). The only requirement is Docker Desktop or Go 1.22+.

One-line installer

# Downloads binary, sets permissions, and starts Callahan
curl -fsSL https://callahanci.com | sh

Docker Compose

git clone https://github.com/goughliam1813/callahan
cd callahan
docker compose up

From source

git clone https://github.com/goughliam1813/callahan
cd callahan/backend
go mod tidy
go run ./cmd/callahan

# In a second terminal:
cd callahan/frontend
npm install
npm run dev

Quickstart

Once installed, open http://localhost:3000 in your browser. You'll see the Callahan dashboard.

1. Click Connect Repository and paste your Git repo URL
2. Callahan auto-detects your language and suggests a pipeline
3. Click Run Build — your first pipeline executes in an ephemeral container
4. Add an LLM API key in Settings to unlock all AI agents

Configuration

Callahan is configured via environment variables. Copy .env.example to .env and fill in your values.

# LLM providers — add at least one
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
GROQ_API_KEY=gsk_...

# Or use local Ollama (no API key needed)
OLLAMA_HOST=http://localhost:11434
DEFAULT_MODEL=llama3

# Storage
DATA_DIR=./data
PORT=8080

Callahanfile.yaml

Pipelines are defined in a Callahanfile.yaml at the root of your repository. The syntax is a superset of GitHub Actions — if you know Actions, you already know 90% of it.

name: my-pipeline
on: [push, pull_request]

jobs:
  build:
    runs-on: callahan:latest
    steps:
      - name: Install
        run: npm ci

      - name: Test
        run: npm test

      - name: Build
        run: npm run build

    # Callahan AI extensions
    ai:
      review: true
      explain-failures: true
      security-scan: true

LLM Setup

Callahan supports any major LLM provider. It tries them in this order: Anthropic → OpenAI → Groq → Ollama. The first one with a valid API key is used.

Pipeline Architect

The Pipeline Architect agent generates a complete Callahanfile.yaml from a plain English description. Access it from the dashboard by clicking Callahan AI in the sidebar.

Example prompt: "Build my Next.js app, run Playwright end-to-end tests, scan with Trivy, and deploy to Fly.io on green PRs."

Build Debugger

When a build step fails, click the AI Explain button next to the failed step. The Debugger reads your full log context and returns a plain-English diagnosis plus a suggested fix.

Code Reviewer

Enable code review in your Callahanfile.yaml with ai.review: true. On every pull request, Callahan posts structured review comments to your Git provider — bugs, security issues, performance concerns.

Supports: GitHub, GitLab, Bitbucket, Gitea.

Security Analyst

Enable with ai.security-scan: true. Runs Trivy (container/dependency scanning), Semgrep (SAST), and gitleaks (secret detection). Findings are explained in plain English. Critical findings block the pipeline.

Secrets

Add secrets in the dashboard under Project → Secrets. They are stored encrypted and injected as environment variables at runtime. Reference them in your pipeline:

- name: Deploy
  run: flyctl deploy
  env:
    FLY_API_TOKEN: ${{ secrets.FLY_API_TOKEN }}

Deploy to Fly.io

- name: Deploy to Fly.io
  uses: callahan/deploy-fly@v1
  with:
    app: my-app-name
    token: ${{ secrets.FLY_API_TOKEN }}

Deploy to Vercel

- name: Deploy to Vercel
  uses: callahan/deploy-vercel@v1
  with:
    token: ${{ secrets.VERCEL_TOKEN }}

Environment Variables Reference

REST API

Callahan exposes a full REST API at http://localhost:8080/api/v1. All endpoints return JSON.

# List all projects
GET  /api/v1/projects

# Trigger a build
POST /api/v1/projects/:id/builds

# Stream build logs (WebSocket)
WS   /api/v1/builds/:id/logs

# AI explain a failed build
POST /api/v1/builds/:id/explain