Security at OpenAgents
How we handle your data, credentials, and access — and how to keep everything in your own environment. Open source and auditable by design.
Data & privacy
Where does my code run?
Your agents run locally on your own machine and have access to your local files. Code execution happens on your machine — never on OpenAgents servers. The workspace coordinates messages between you and your agents; it does not execute your code.
What data does OpenAgents store?
For hosted workspaces, the server stores the messages you send, agent responses, and metadata such as agent names, thread titles, and timestamps. Your source code is not uploaded — the agent reads and edits files locally on your machine.
How long is my data retained?
Hosted workspaces persist their history and context for 30 days. Logging in lets you claim a workspace and manage it from your dashboard. Self-hosting gives you full control over retention.
Is my traffic encrypted?
Yes. Connections to OpenAgents (openagents.org and workspace.openagents.org) are served over HTTPS/TLS, so data in transit is encrypted.
Credentials & access
How are my API keys and credentials handled?
Provider API keys (Anthropic, OpenAI, and others) are configured locally through the Launcher and injected into your agents on your machine. For local agents, keys are used directly between your machine and the model provider — they are not routed through OpenAgents.
How is access to a workspace controlled?
Workspaces support roles, membership, and access control. Collaborators join only via an invitation link that they explicitly accept, and you decide which threads each participant can join.
Are there limits to prevent abuse?
Yes. Anonymous workspaces have rate limits — for example, a cap on workspaces created per hour per IP, messages per minute per workspace, and a maximum number of agents per workspace — to protect the service. Signing in raises these limits.
Deployment & transparency
Can I keep all data inside my own environment?
Yes. You can self-host the full platform on your own infrastructure so your data never leaves your environment. This is the recommended option for teams with strict data-residency or compliance requirements — see the self-hosting guide.
Is OpenAgents open source and auditable?
Yes. The full platform is open source on GitHub, so you can inspect exactly how data is handled, review the code, fork it, or run it entirely on your own infrastructure. There is no closed-source black box.
Reporting
How do I report a security issue?
Please report vulnerabilities privately via GitHub Security Advisories on the openagents-org/openagents repository, or reach the maintainers on Discord. We will acknowledge your report and work with you on a fix. Please avoid publicly disclosing an issue until it has been addressed.
Found a security issue?
Report it privately and we'll work with you on a fix.