Ellipsis

Ellipsis is a managed cloud platform for building and running coding agents for feature generation, background tasks, and code review. Teams define agents in YAML, deploy them by merging changes, and trigger runs through API calls, schedules, or events. Each session uses an isolated cloud sandbox prepared with repositories, dependencies, and tools. The platform supports Claude Code and Codex, and says teams can switch agents and models without rebuilding an agent.

Features include triggers, durable conversations, REST API and CLI access, structured output, shared skills, templates, session search, and live steering. Listed integrations include GitHub, Linear, Slack, and Sentry; Jira is marked coming soon. Teams can set repository and read/write permissions, while session-created GitHub credentials are revoked when a sandbox ends. Code is kept in the sandbox and deleted with it; action logs and configurable session data retention are also listed. Ellipsis supports spending caps per session, day, and month. Its Individuals plan costs nothing with a Claude Code or Codex subscription. Organizations pay tokens plus a 10% platform fee, with CPU and memory billed by usage.

Ellipsis provides configurable agents, isolated sessions, integrations, and controls for access and spending. Individuals need a Claude Code or Codex subscription for the free plan, while organizational charges depend on tokens and usage-based compute.

Ellipsis suits engineering teams building their own coding agents, workflows, and agent-powered applications. It may also suit individuals with a Claude Code or Codex subscription who want to build and run agents.

Ellipsis is a cloud platform for building and running coding agents across feature work, background tasks, and code review. It suits engineering teams creating their own agent-powered workflows. Its strongest case is configurable automation with controls for access, data retention, and spending; its main trade-off is usage-based organizational billing.

Overview

Ellipsis gives teams a managed place to configure agents and connect them to repositories and engineering tools. YAML configuration, deployment through merged changes, and event or schedule triggers suit teams that want agent workflows managed alongside their code. It is less compelling for someone seeking a ready-made coding assistant for individual use: the free plan requires a Claude Code or Codex subscription, and the platform is oriented toward building workflows.

Key features

Agents can start through API calls, schedules, or events. Each run receives an isolated cloud sandbox with repositories, dependencies, and tools prepared for it. Support for Claude Code and Codex, with agents and models switchable without rebuilding the agent, gives teams room to change their underlying setup as needs evolve.

REST API and CLI access, triggers, durable conversations, structured output, shared skills, templates, session search, and live steering support both automated runs and intervention while an agent is working. Integrations include GitHub, Linear, Slack, and Sentry; Jira is marked coming soon. The breadth is useful for teams connecting agent work to existing engineering processes, though teams relying on Jira will need another route for now.

Repository access and read or write permissions can be scoped. GitHub credentials created for a session are revoked when its sandbox ends, and code in the sandbox is deleted with it. Teams can configure session data retention from zero hours to one year, while agent actions are logged. Ellipsis says it is SOC 2 Type I certified, with Type II in progress. Spend caps can be set per session, day, and month, and are checked before a sandbox is created. Together these controls make unattended agent work more manageable for teams with security and budget requirements.

Pricing

Individuals: free (billed Free with a Claude Code or Codex subscription). This gives individuals a no-cost way to use the platform if they already subscribe to one of those tools; that subscription is a prerequisite.

Organizations: custom pricing. Managed SaaS charges tokens at cost plus a 10% platform fee, along with $0.142 / vCPU-hour and $0.024 / GB-hour. Your AWS Account runs in your AWS VPC, with tokens and compute billed by AWS and a 10% platform fee on token cost. Both options include the full feature set, have no per-seat fees, and charge no idle fees. Usage-based compute and token charges mean costs depend on activity, so the spend caps matter.

Support is separately priced monthly: Standard Support is $5,000 a month and includes Slack support, guided onboarding, help building custom agents, self-guided BYOC, a 48-hour first response, and a shared support engineer. Advanced Support is $10,000 a month; it adds BYOC deployment help, security questionnaires and procurement support, and a 24-hour first response. Premier Support is $15,000 a month; it adds priority feature requests, a four-hour first response, and a named support engineer. These packages are aimed at organizations that need direct support; the platform otherwise is self-serve without a support package.

Platforms

Ellipsis is available through web and API access. It supports GitHub repositories and offers CLI access for working with the platform.

Who it's for

Ellipsis is best suited to engineering teams building custom tools, workflows, and applications around coding agents. Teams that want configurable triggers, isolated sessions, repository permissions, and usage controls get a broad foundation without per-seat platform fees. Individuals can start free with a Claude Code or Codex subscription, but teams seeking a turnkey assistant or predictable flat organizational pricing may prefer another option.

Pros and cons

• Pros: YAML-defined agents deploy through merged changes, which lets teams manage configurations in their code workflow.
• Pros: Isolated sessions, scoped repository permissions, revocable session credentials, configurable retention, and spend caps provide practical controls for automated work.
• Pros: Full platform access without feature tiers or per-seat fees gives organizations access to the same capabilities as they scale seats.
• Cons: Organizational token and compute charges vary with use, making costs less predictable than a fixed subscription.
• Cons: Individuals need an existing Claude Code or Codex subscription to use the free plan.
• Cons: Direct support requires a separate package priced from $5,000 a month, a substantial commitment for teams that need help.

Alternatives

CodeRabbit is worth considering for teams looking for a freemium code review tool with a free tier for public repository reviews forever and support across API, extensions, desktop platforms, web, and self-hosting.

Gitar offers a free-for-open-source option for qualifying public GitHub or GitLab repositories with an OSI-approved license; it may fit open-source maintainers who want Pro plan features at no cost.

PR Quorum has a free tier capped at 50 reviews per month, two repositories, and three inline comments per review, making it a more bounded option for teams with modest review volume.

AI Code Reviewer is free on Linux and self-hosted, with users paying only their model provider's usage; consider it when self-hosting and avoiding an additional platform charge are priorities.

Codacy offers a free developer plan with an IDE plugin, support for 38 programming languages, and integrations for VS Code, JetBrains, and Cursor, which may suit developers focused on IDE-based analysis.

CodeAnt AI offers a 14-day free trial with no credit card required, useful for a time-limited evaluation before committing to a paid plan.

CodeKang has a forever-free tier for five private repositories with basic AI summaries, a fit for teams that want those specific capabilities at no cost.

Cubic offers 20 PR reviews a month on its free Starter plan, with free use for public repositories, a straightforward alternative for teams with limited review needs.

Browse AI Code Review Tools, Code review software, and AI Coding Assistants to compare more options.

Verdict

Choose Ellipsis if your engineering team wants to build and operate custom coding-agent workflows with isolated sessions, integration options, and explicit permission and spending controls. The full platform without per-seat fees is a strong fit for teams whose usage justifies a managed foundation. Look elsewhere if you need fixed organizational pricing, direct support without a costly package, or an individual tool that works without a Claude Code or Codex subscription.