---

🧠 Drupal DevOps Co-Pilot

An open-source AI assistant that automates the full Drupal DevOps lifecycle — from Jira issue to GitLab merge request to EC2 deployment and QA.

AI-powered agent that plans, writes, commits, opens MRs, runs Drush QA, and deploys Drupal changes.

Built with FastAPI, GitLab CI, Jira API, and local or cloud LLMs like Ollama (Qwen2.5-Coder) or OpenAI-compatible endpoints.

---

✨ Features

• Reads context: Jira ticket + repo scan + guardrails
• Generates code: unified diff or JSON file manifest (for scaffolds)
• Sanitizes safely: fixes JSON escapes, unescapes PHP $vars, removes rogue declare(strict_types)
• Commits & pushes: clean messages, opens Merge Request automatically
• Runs QA: Drush (cr, pm:install, etc.) on EC2 or your target host
• Reports back: comments/updates to MR and Jira (optional)

Note

From ticket → deployment in minutes, consistently.

---

🧱 Architecture (High-level)

• Agent Brain (Python) → prompt builder → LLM call → patch/manifest → sanitizers → Git ops
• LLM Runtime → Ollama (local) or any OpenAI-compatible /v1 endpoint
• Integrations → GitLab (MRs), Jira (optional), EC2/remote host (Drush QA & deploy)

System Architecture:
End-to-end automation from Jira → GitLab → EC2 using FastAPI and Ollama.

---

🛠️ Requirements

• OS: macOS or Linux (Apple Silicon & Intel both OK)
• Python: 3.10+
• Git: 2.30+
• Ollama (for local LLM) or an OpenAI-compatible endpoint
• GitLab repo (personal access token if creating MRs via API)
• Drupal project to operate on (monorepo or separate checked out under work/)

---

📎 TL;DR

# Install deps
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
ollama pull qwen2.5-coder:14b-instruct-q4_K_M

# Configure .env (see above), clone your Drupal repo to work/drupal-project

# Run
copilot-one-shot CCS-128
# or
copilot-workflow CCS-128 && copilot-ai-review-merge "<MR_URL>" --auto-merge --deploy

If you get stuck or want to propose an agent skill, open an issue — or ping me.

---

🛣️ Roadmap & ideas for v2+

Feature	Status
Multi-agent orchestration	🚧 In progress
DeepSeek/StarCoder LLMs	🧩 Planned
Plugin system	🧠 Concept
Web UI dashboard	🧭 Beta
Provider auto-benchmarking	💡 Idea

---

🛡️ Reliability

• Strict path guardrails (web/modules/custom/)
• JSON repair for invalid escapes (\Drupal, $vars)
• PHP sanitizer removes declare(strict_types) and rogue define()
• Retry logic for Ollama timeouts
• Fallback to manifest mode if patch fails

---

🤝 Contributing

We welcome ideas and pull requests!
See CONTRIBUTING.md for guidelines.

🧭 Code of Conduct

Please review our CODE_OF_CONDUCT.md.
We aim for a respectful, inclusive, and helpful space for all contributors.

---

📚 Documentation

• 🚀 Quickstart (local)
  Clone, setup venv, install deps, configure .env, run smoke test.

• 🐳 Docker Quickstart (public images)
  One command to pull & run API, Web, and Ollama.

• 🛠️ Docker for Development
  Live-reload containers, source mounts, and developer workflow.

• 🖥️ Local LLM (Ollama) tips
  Installation, model management, tuning parameters, and troubleshooting.

• 🛠️ Configuration Guide
  Environment variables, LLM providers, and GitLab/Jira setup.

• 🧰 CLI reference
  Available commands, flags, and usage examples.

• 📊 Architecture Overview
  High-level components, data flow, and integrations.

• ⚙️ Flow Overview
  End-to-end pipeline: Auto-Dev → AI Review → Deploy → QA Verify.

• 💻 Working in a New Terminal Session
  How to activate the environment, run backend/frontend locally, and test commands.

• 🔧 Having issues? See Troubleshooting Guide
  Common issues, debugging tips, and FAQs.

• 🛡️ Security Policy
  Responsible disclosure process, supported versions, and scope.

• 📄 Contributing Guide
  How to contribute, commit message conventions, and project goals.

---

🙌 Acknowledgements

Powered by the open-source communities behind Drupal, FastAPI, and Ollama.

---

💬 Questions or ideas?

Join the discussion on GitHub Discussions —
we’d love to hear from contributors, Drupal engineers, and AI enthusiasts!