Appearance
Connectors overview
Connectors are how Liddr learns about your organization. Each one is a hand-coded, first-class integration with a source system — there are no generic "build your own" templates. Adding a connector is a guided flow inside Settings → Connections, where each source renders its own card and credential fields.
Liddr has two distinct kinds of connector, and the difference between them is not cosmetic — it is the core of how Liddr keeps your data safe.
The access principle
The organization knowledge graph has no per-user or per-role access control. Anything written into it is readable by every member of your organization through Know. That single fact decides where every connector belongs:
- Org-shared, non-sensitive data — structure, business rules, work artifacts, and authored team knowledge — is ingested into the shared knowledge graph.
- Personal, record-level, or sensitive data — CRM records, support tickets, HR data, mailboxes, files, chat — is never brought into the graph. It is reached live, per user, through a personal connector when you ask.
Why this matters
If a source contains rows that not everyone should see (a deal in your CRM, a support ticket, a salary in an HR system, someone's inbox), it can't be brought into the shared graph — there's no way to filter it per viewer. Those sources are reached through personal connectors that run under each user's own credentials. This is why HubSpot, Zendesk, Workday, and Gmail are personal connectors, not shared ones.
What is shared vs. personal
| Data | How it's reached | Why |
|---|---|---|
| Repos, issues, PRs, work items (GitHub, GitLab, Jira, Linear…) | Shared | Shared work artifacts and structure |
| Incidents, alerts, error signals (PagerDuty, Opsgenie, Sentry…) | Shared | Org-wide operational signal, not personal |
| Docs, wikis, knowledge bases (Notion, Confluence, GitBook, Guru…) | Shared | Authored team knowledge meant to be shared |
| Pure structure (Databricks jobs, SQL warehouses) | Shared | Metadata/definitions, not row-level data |
| CRM records (HubSpot, Affinity) | Personal | Record-level business data; not everyone should see every deal |
| Support tickets (Zendesk, Freshdesk, Intercom, Front) | Personal | Customer records, often sensitive |
| HR / recruiting (Workday, BambooHR, Greenhouse) | Personal | Personal and confidential |
| Files, chat, mailboxes, calendars (Slack, Box, Dropbox, Gmail, Drive…) | Personal | Per-user data with personal scope |
INFO
For sources like a CRM or BI tool, the structure and business rules could in principle be shared, but the data rows cannot — so those systems stay personal-only for now.
Shared connectors
Shared connectors bring org-shared content into the knowledge graph and keep it current. There are roughly two dozen of them, spanning developer and project tracking, docs and knowledge, and incident and ops signals. Each tags its content with its own source, so Know, Signal, and Forge can reason across all of them together.
See Available connectors for the full catalog.
Personal connectors
Personal connectors are set up per user. They aren't brought into the graph — Liddr reaches the source live, under your own credentials, only when one of your questions needs it. Once you connect one, Know can use it during a conversation to fetch your personal or record-level data.
See Personal connectors for the available connectors and how Know uses them.
Where connectors live
All connectors are managed from Settings → Connections. Choosing Add connector opens the picker right inside Connections. Shared connectors are connected once for the whole organization; personal connectors are added per user from the catalog.
The lifecycle at a glance
Every shared connector moves through the same steps:
- Add connector — pick the source and enter its credentials.
- Test — verify the credentials. A connector is not considered connected until you have tested it.
- Sync — once connected, it brings content into the graph and stays current, with progress shown as it runs.
- Edit — update credentials or settings later from the connected card.
See Connecting sources for the step-by-step.