Integrations
Connect your existing tools to build the knowledge graph automatically. Cograph uses read-only access to analyze signals from your data sources — nothing is stored that isn't needed for expertise mapping.
Privacy note: Cograph never reads message content or code content. It analyzes metadata: who creates what, who collaborates with whom, and in which topic areas. No PII is sent to AI APIs.
Google Workspace
The foundational integration. Google Workspace syncs your employee directory (the source of truth for who is in your org), org chart, and Google Docs authorship signals.
What gets synced
Permissions required
admin.directory.user.readonly— read user profiles and directoryadmin.directory.group.readonly— read group membershipsdrive.metadata.readonly— read file metadata (not content)
Setup
- Go to Settings → Integrations → Google Workspace
- Click Connect — you'll be redirected to Google OAuth
- Sign in with a Google Workspace Admin account
- Grant the requested permissions
- The initial sync runs automatically (10–30 min for large orgs)
A Google Workspace Admin account is required to authorize the integration. The connection is specific to your Google Workspace domain.
Slack
Analyzes public Slack channel activity to build collaboration networks and identify communication-based expertise signals. No message content is stored.
What gets synced
Permissions required
channels:read— list public channelsusers:read— read user profileschannels:history— read message metadata (not content)groups:read— list private channels (optional)
Setup
- Go to Settings → Integrations → Slack
- Click Connect — you'll be redirected to Slack OAuth
- Select the workspace to connect
- Grant the requested permissions
- Choose which channels to include in signal collection
Only public channels are analyzed by default. Private channel analysis requires an additional permission grant and is opt-in per channel.
GitHub
Maps code ownership across repositories by analyzing commit history and pull request patterns. Provides the strongest signal for technical expertise scoring.
What gets synced
Permissions required
contents:read— read repository contents and commitspull_requests:read— read PR history and reviewsmetadata:read— read repository metadata
Setup
- Go to Settings → Integrations → GitHub
- Click Connect — you'll be redirected to install the GitHub App
- Select your GitHub organization or specific repositories
- Grant access to the repos you want analyzed
- Cograph creates account links between GitHub usernames and Cograph employees
Uses a GitHub App (not OAuth) for fine-grained repository-level access control. You can choose which repositories to include.
Sync Schedule
Cograph uses incremental syncing — only new activity is processed after the initial backfill.
Full historical backfill — all commits, documents, messages
Incremental — only new activity since last sync
Forces an immediate incremental sync
Full recalculation of all expertise scores and relationships
Large organizations: Initial syncs for organizations with 500+ employees or 100+ repositories may take several hours. The graph will be incomplete until the initial sync finishes.
Troubleshooting
The integration shows "Error" status
OAuth tokens expire after 60 days. Go to Settings → Integrations and click Reconnect to reauthorize. This is expected behavior — we rotate tokens aggressively for security.
Some employees are missing from the knowledge graph
Account linking connects GitHub usernames and Slack handles to Cograph employee records via email matching. If an employee uses a different email for GitHub, go to Settings → Account Links and add the mapping manually.
Expertise scores seem too low / inaccurate
Scores improve as more integrations are connected. Connecting only GitHub will give technical expertise but miss documentation and communication signals. Connect all three integrations for the best results. Scores also take 1–2 weeks to stabilize as historical data is processed.