Skip to main content
Open Source

Open-source AI agent
security plugins

We offer locally run, open-source security plugins for OpenClaw and DeepSeek Harness. They help developers identify risks, record security events, and alert on or control high-risk actions according to policy while agents run.

Each plugin has its own source repository. Both share a feedback form and technical discussion group.
OpenClawRuntime security plugins
DeepSeek HarnessRuntime security plugins
Runs locally
Open source for review
Plugins

Choose a compatible security plugin

Visit the OpenClaw and DeepSeek Harness plugin repositories for installation instructions, and use the shared form to report issues.

Usage modes

From a local plugin to centralized security management

The open-source plugins provide local security capabilities; the enterprise platform supports centralized governance across agents. Their scopes are distinct.

01 / LOCAL

Plugin mode

Install and configure a plugin locally to inspect a single agent runtime. See the corresponding repository for requirements and instructions.

02 / MANAGED

Platform mode

The Agent Runtime Security Governance System provides agent inventory, collector event intake, and policy management. See each repository for how its plugin connects to the platform.

03 / PRIVACY

Privacy and data

Plugin settings determine how data is stored locally and reported. Before connecting to the platform, review the event reporting scope and related settings.

Open-source scope:Both plugins use the Apache-2.0 license. The enterprise platform is outside the plugins’ open-source scope; see each repository’s license for covered files.
Support & Community

Get support and join the conversation

Both plugins share one online form for general usage questions, bugs, and suggestions, and one official WeChat group for installation, configuration, and use discussions.

Official feedback channel

Use the form below for usage questions, bugs, and suggestions about either plugin. Name the relevant plugin in your description.

For security vulnerabilities, use the form above and mark the report “Security vulnerability”. Do not disclose details in the WeChat group or a public issue.