Introduction

Ellipsis runs coding agents in the cloud. Commit one environment YAML to your repo, then spawn the same Claude Code or Codex you run locally with the agent CLI: each session in its own sandbox, in parallel, recorded, under budgets you set.

environment:  repositories:    - name: api-repo  variables:    - name: DOPPLER_TOKEN

Why use Cloud Agents?

• A laptop runs one agent while you watch it. The cloud runs a fleet, each session in its own sandbox, and keeps working after the lid closes.
• Agents start themselves: GitHub, Slack, Linear, and Sentry events, cron schedules, and @ellipsis mentions all spawn sessions.
• Nothing lives on one engineer's machine. The environment is a YAML file in the repo, so every teammate and every agent runs the same setup.

How Ellipsis helps

• Scoped permissions: each session's GitHub token carries exactly the access you grant and expires with the session.
• Hard budgets: caps per session, per agent, and per developer, enforced before a sandbox is created.
• Full records: every session is searchable and exportable. "Which session produced PR #482" is a query, months later.
• Fast starts: dependency installs bake into a cached image, so repeat sessions start in seconds.

One YAML updates the whole team

Add an environment variable or an MCP server as one reviewed line; every agent picks it up on merge.

 ellipsis:
   version: v1
   name: Standup digest
 
 environment:
   repositories:
     - name: api-repo
   variables:
+    - name: DOPPLER_TOKEN
 
+mcp_servers:
+  - slack

Use cases

Surfaces

Integrations

GitHub

Repos, tokens, events, and the @ellipsis mention.

Slack

Mention @ellipsis in a thread; agents answer with code context.

Linear

Agents implement issues and reply on tickets.

Sentry

Alerts spawn root-cause investigations.