RepoBird

About RepoBird

RepoBird runs one-shot cloud coding agents against your repositories: you describe a task in the dashboard or the CLI, an agent works in an isolated sandbox, and what comes back is a pull request with a summary, tests and the changed files attached — prompt to PR, with no chat session to babysit. The design choice that distinguishes it is parallelism. Multiple agents run concurrently through a run queue, each on its own issue, so a backlog of small fixes clears simultaneously rather than one conversation at a time; the concurrency ceiling is what the paid tiers actually sell, rising from one active run on Free to eight on Starter and sixteen on Pro, with thirty queued on each. It connects through a GitHub App or a GitLab button, works on private repositories, and supports bring-your-own-key so you can point it at your own model provider and keep that billing separate from the platform's. A template marketplace ships more than 350 pre-built agentic prompts — unit test generation, README generation, an OWASP-flavoured security audit, dependency updates, lint fixes, DRY refactoring — each with a rating and an execution count, so you can start from something other people have already run rather than authoring the prompt yourself. Every run uses an approved credit cap agreed before it starts, so a runaway agent cannot quietly spend the month.

Key Features

Tags