Ellipsis

Ellipsis is a code modifier tool. A managed platform that runs coding agents defined in YAML, covering pull request code review, cloud software factories, and API triggered background agents, with per session credentials, budget caps, and audit logs.

Ellipsis is a managed platform for running coding agents that you define yourself. Instead of opening a terminal on your laptop, you describe an agent in a YAML file stored in your repository, and Ellipsis executes it on cloud virtual machines. The configuration covers the prompt, the repositories the agent can touch, setup hooks, compute size, and spending budgets. You pick the harness and the model, and the platform supplies the infrastructure, permissions, and logs around it. The homepage leans on three standing use cases. Software factories define every coding agent in YAML and run them in the cloud. Code review agents patrol each push, with reviewers you define checking incoming changes. Background agents launch from a single API call, which suits teams that want to trigger work from their own tooling rather than from a chat window. The supported harnesses include Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot, Cursor, Pi, Grok Code, and Antigravity, so an existing subscription or preferred agent does not have to be abandoned. Switching between backends such as Claude Code and Codex is a one block edit in the config. The audience is engineering organizations rather than solo hobbyists: teams running several repositories, teams that care about who authorized what, and teams that need cost ceilings before an agent starts burning tokens. Budgets can be capped per session, day, week, and month, and a session that hits its limit is stopped. Governance is the pitch. Credentials are minted per session, scoped to allowed repositories, and revoked when the sandbox ends. Source code is not retained beyond the session, while a full audit trail of commands, file changes, and tool calls survives the sandbox. Against alternatives, Ellipsis is closer to a control plane than to a single review bot. General purpose code review services give you an opinionated reviewer with fixed behavior. Ellipsis gives you a configurable runner that can also be steered toward issue triage, test coverage checks, or CI fixes. The tradeoff is that it expects a team already comfortable writing agent configs and managing repository permissions.