Our take
Activeregular commits- Activity
- 53
- Community
- 53
- Issues
- 46
- Pull requests
- 53
+127 stars in the last 30 days
Open-Inspect is an actively developed, broad background-agent system, but its single-tenant security model and concentrated contributor activity make deployment boundaries and project continuity important considerations.
Good fit if
- Your organization wants background coding sessions connected to web, Slack, GitHub, Linear, or automation workflows.
- You can deploy for one trusted organization and configure access, repository scope, and sandbox infrastructure accordingly.
- You are comfortable adopting a TypeScript project that is actively receiving commits and merged pull requests, despite having no release in the past year.
Look elsewhere if
- You need strong tenant isolation for unrelated customers. The project explicitly says it is designed for single-tenant deployment.
- Your adoption policy requires development to be spread across multiple contributors. Most measured contribution activity comes from one person.
- You need a project with a recent published release or measured recent issue closure activity.
All health signals
| Last commit | 2026-10-30 (today) |
|---|---|
| Commits, last 90 days | 100+ |
| Releases, last 12 months | 0 |
| Contributors | 54 (top contributor: 80% of commits) |
| Issues closed, last 90 days | 1+ |
| Pull requests merged, last 90 days | 39 (typically merged in 1 day) |
| Project age | 9 months |
Checked on 2026-10-30 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
3,490Forks
508Change since Nov 14, 2026: +2 stars·+1 forks
Overview
Open-Inspect is an open-source background coding agent system inspired by Ramp’s Inspect. Its control plane uses Cloudflare Workers and Durable Objects, while sandbox sessions run coding-agent harnesses such as OpenCode and Anthropic’s Claude Agent SDK. It gives developers a full development environment in which agents can work on code tasks without occupying the developer’s active session.
Teams can start and collaborate on sessions through the web UI, Slack, GitHub, Linear, or webhooks. The project is designed for single-tenant deployment among trusted members of one organization, rather than isolation between unrelated customers.
Key Features
- Run coding tasks in isolated sandboxes with Node.js, Python, git, browser automation, and optional browser-based VS Code.
- Start sessions from the web UI, Slack, GitHub PRs, Linear issues, or authenticated webhooks.
- Collaborate in multiplayer sessions, with prompts attributed to their authors in git commits.
- Work across multiple repositories in one session, or save repository sets as environments.
- Run scheduled and event-triggered automations, including cron, GitHub workflow, Sentry, and webhook triggers.
- Spawn parallel child sessions in separate sandboxes for independent subtasks.
- Choose among supported AI model providers and agent harnesses, with per-session reasoning controls.
- Integrate sandbox infrastructure through packages for Modal, Daytona, E2B, and Boat.
Use Cases
- A development team can delegate implementation or maintenance work to agents while continuing other work, then review the resulting pull requests.
- Teams that coordinate work through Slack, GitHub, or Linear can launch coding sessions from those existing workflows.
- An engineering team can automate recurring repository tasks or start agent sessions in response to GitHub workflow runs or Sentry alerts.
- Developers working across related repositories can use a shared environment for coordinated changes and separate pull requests.
What you need
Detected in the repository
- Node.js >=24.0.0 (from package.json)
- A Compose file to run it with its related services
- Automated checks on GitHub Actions
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
Start with the setup guide for local, contributor, and deployment paths. The repository also provides deployment instructions and an architecture guide. See the repository README for full details.
Considerations
- The README warns that this is single-tenant software: team membership and access controls do not provide isolation between separate customer tenants.
- Repository access depends on the shared GitHub App installation and session or team grants. Credentials are cached on disk, snapshots can retain them, and removing a grant does not immediately revoke already-issued tokens. GitLab uses a deployment-wide personal access token rather than a per-session token.
- The project requires Node.js 24 or later and uses Docker Compose. The supplied requirements data reports CI but no tests.
- Development is active, with regular recent commits and merged pull requests, but there have been no releases in the past year. Measured contributor activity is concentrated mostly in one person, while recent issue closure activity is limited.
- The MIT license permits broad use, subject to its license terms.
Found this useful?
Share it with someone who would like background-agents.