Skip to main content
All three integrations are optional. You can use local findings, the catalog, HTTP and MCP without them. Complete provider connection and selection in Settings; credentials alone do not select a destination.

GitHub

Use an operator-owned GitHub App and select its installation/repository. Basic issue linking uses Issues read/write; linked PR status needs Pull requests read. GitHub access uses short-lived scoped installation tokens, not a stored personal access token. The configured replay/advisory PR gate additionally needs Checks read/write, Pull requests read/write and Contents read, signed webhook deliveries and a suitable preview URL. Candidate inventory discovery is a separate capability. Actions-based pilot execution has its own trusted adapter and permissions. Enable only what your deployment uses.

Linear

Connect workspace OAuth (actor=app) and select the intended team. This is not a personal API-key integration. Issue creation attributes the initiating user; finding assignment can synchronize by member email, including an assignment made before issue creation. Existing installations may require reconnection for issue-write permissions.

Slack

Connect the workspace, select a channel and invite the bot. Required scopes include chat:write, channels:read and groups:read. Accepted findings can notify that channel; notifySlack: false suppresses notification for an ordinary individual submission. Review failures in logs rather than assuming a stored finding was delivered.

Configure your own deployment

Direct Node integrations use deployment-owned client IDs/secrets and the configured public origin. Hosted broker deployments keep provider app credentials at the broker, with tenant-owned workspace/repository selection. Use configuration for variable names. Register callbacks on your own HTTPS origin; never copy a private deployment’s identity. Grant additional permissions only after reviewing the capability that needs them. Keep secrets outside source files and documentation examples.