How does Orquesta work?

You connect your development tools (GitHub, Vercel, etc.) once, then your team writes prompts describing features they want to build. Our AI agent runs on your machine, reads your codebase, writes code, creates PRs, and deploys - all while you watch in real-time.

What AI models do you support?

We always support the latest Claude and GPT models, plus local models via Ollama or vLLM. With Orquesta CLI, you can configure multiple LLM endpoints and sync them across your team from the dashboard.

Yes. The AI agent runs on YOUR machine, not ours. Your code stays local. Credentials are encrypted with AES-256-GCM. We use row-level security in our database. Your secrets are never stored in plain text.

Can I use my existing GitHub workflow?

Absolutely. Orquesta creates standard git branches and pull requests. Your existing CI/CD pipelines, code review process, and GitHub Actions all work exactly as before.

What happens if the AI makes a mistake?

Every change is a git commit, so you can always revert. You review PRs before merging, just like human code. Plus, you can watch the AI work in real-time and stop it anytime.

What is Orquesta CLI?

Orquesta CLI is a standalone coding assistant that runs locally. Configure LLM endpoints (OpenAI, Anthropic, Ollama), sync configurations from the dashboard, and track prompts across your team. Install with: npm install -g orquesta-cli

Do I need to be technical to use Orquesta?

Non-technical team members can write prompts, but you'll want at least one developer to review the generated code. That's actually the point - anyone can contribute ideas, developers ensure quality.

Git-Native AI Development: Traceability in Every Commit

Explore how Orquesta's git-native AI approach ensures accountability and easy rollback by making every AI-driven code change a real commit with traceable history.

When an AI agent writes code, the stakes are high. Unlike human developers, AI lacks intuition about context and intent. This is why traceability becomes crucial, especially when AI-driven changes are involved. At Orquesta, we make every AI action a real git commit, ensuring full accountability and easy rollback. Let’s explore how this fundamentally changes the landscape of AI-assisted development.

The Importance of Traceability in AI Code Development

Traceability in software development means being able to track every change, who made it, and why it was made. In human-driven development, this is often achieved through clear commit messages and detailed logs. However, when an AI agent takes the helm, the need for traceability intensifies.

Why Every Action as a Commit Matters

Accountability: Each commit includes an author, timestamp, and a diff. This ensures every change can be traced back to a specific prompt and decision.
Auditability: Full audit trails mean any AI-generated change is recorded and can be reviewed. This includes the prompt, the resulting code changes, and any cost implications.
Rollback: If an AI-driven change introduces a bug or undesired effect, reverting is as simple as rolling back a commit.

Orquesta's Git-Native Approach

We designed Orquesta with git-native principles at its core, making sure that every action taken by the AI agent is recorded as a git commit. Here's how it works:

Agent Execution: Whether in Auto, SSH, Agent, or Batuta mode, each AI action is captured as a commit. This includes any code changes, refactorings, or integrations.
CLAUDE.md Synchronization: Our coding standards are enforced every time the agent executes, and results are synced with CLAUDE.md, ensuring that the AI adheres to predefined coding guidelines.
Real-time Streaming: Every line of output is streamed live, allowing team members to monitor the AI agent's actions as they happen.

Real Workflows with Commit Traceability

Consider a scenario where the AI is tasked with refactoring a critical piece of the codebase for performance optimization. With Orquesta:

Prompt Submission: A team member submits a refactor prompt through the platform or directly via the Telegram bot.
AI Execution: The AI agent executes the task, makes the necessary code changes, and commits them to the git repository.
Commit Details: Each commit is tagged with the AI agent as the author, includes a timestamp, and provides a detailed diff of changes.
Team Lead Approval: These changes are subjected to quality gates where a team lead can review, simulate the changes, and sign off before they are merged into the main branch.
Rollback and Audit: If needed, any of these changes can be rolled back seamlessly, and a full audit trail is available for review.

Enhancing Team Collaboration

Traceability isn't just about control; it’s also about collaboration. By using git as the single source of truth, teams can:

Collaborate Across Geographies: Git-native commits mean that even distributed teams have a consistent view of the changes, fostering greater collaboration.
Role-Based Permissions: With Orquesta, teams can define who can submit prompts, who can approve changes, and who can execute them.

The Future of AI in Development

AI is transforming the way we think about code, but without proper traceability, we risk losing control over our software landscapes. By making every AI action a real git commit, Orquesta provides a robust framework for accountability and collaboration.

Our approach ensures that AI can be a trusted partner in development, providing both speed and security. This is not just a feature—it’s a necessity in the evolving world of AI development.

Conclusion

Incorporating AI into development workflows brings immense potential, but also challenges. Traceability through git-native commits is a linchpin for integrating AI responsibly. With Orquesta, you can harness AI's power without sacrificing control, knowing that every action is accountable and reversible.

Traceability should not be an afterthought but a fundamental component of AI-driven development. Orquesta stands by this principle, ensuring that every AI action is a part of the project's history, ready to be revisited, audited, or rolled back whenever necessary.