Agent

Describe it. Get a pull request.

Describe what you want changed in plain language. The agent reads your repo, plans the work, writes code, and opens a PR — from any device.

Autonomous coding, start to finish

Describe what you want in plain language. The agent reads your repo, plans changes, writes code, and opens a pull request without manual steps.

Works directly with the GitHub API. Branches, commits, and PRs are created in your repo — no proprietary platform lock-in.

A mobile-friendly chat UI lets you create tasks, follow progress in real time, and send follow-up instructions from any device.

Trigger the agent from GitHub issues with a simple /a21e command. Results are posted back as comments with PR links.

Pay only for what you use with dynamic credit pricing based on instruction depth, repository size, and model workload.

If an AI call fails, credits for that step are automatically refunded. The base fee covers orchestration — you only pay for work completed.

Install the GitHub App and select which repositories the agent can access.

Tell the agent what you want in natural language — a bug fix, a new feature, a refactor, or anything else.

The agent reads your file tree, plans the changes, fetches relevant files, and generates new code.

A new branch is created, changes are committed, and a pull request is opened — ready for your review.

Pricing

Dynamic credits per task

Credits scale with repository complexity, instruction depth, and model effort per call.

Give natural-language instructions and get pull requests back. No IDE, no terminal, no setup.