Linear connector

Index Linear issues, projects, and cycles. Two flavours: Linear OAuth (recommended for workspaces) and Workspace API key (fastest for solo / small teams).

Option A — Linear OAuth (recommended)

  1. In the Gigamcp admin console, open Connectors → Add → Linear.
  2. Click Connect to Linear. You will be bounced to Linear's OAuth consent screen. Approve the read-only read scope.
  3. Linear returns you to /api/oauth/linear/callback; the credential is stored KMS-encrypted in the per-tenant connector_credentials row.
  4. The connector status flips to Active once the first sync finishes (~1-5 min depending on workspace size).

Rotation: the OAuth refresh-token flow keeps the access token fresh automatically. Revoking the install in Linear (workspace settings → API → applications) invalidates the credential immediately.

Option B — Workspace API key

  1. On Linear: Workspace settings → API → Personal API keys → New key. Copy the lin_api_… token.
  2. In the Gigamcp admin console, paste the token under Connectors → Add → Linear → workspace API key.
  3. Click Test connection. We list a few teams to confirm access.

Team allowlist

The optional allowlist constrains which Linear teams the connector exposes to MCP tools and the RAG crawler. Empty = every team visible to the credential. Format accepts either team keys (the short prefix used in issue identifiers, e.g. ENG for ENG-123) or full team UUIDs.

What gets indexed

  • Issue titles, descriptions, comments, labels, assignees, state transitions.
  • Project descriptions + content blocks.
  • Cycle metadata (start / end dates, completion progress).
  • Issue → project / cycle relationships, so RAG can join queries across them.

What does not get indexed

  • Files / attachments uploaded to issues.
  • Teams the credential cannot read (Linear teams have per-team membership; the connector inherits the credential's view).
  • Issues older than the per-team cap (default 2000 most-recent issues per team — tunable post-launch).

Audience visibility

Each team is configured as a separate knowledge source with its own audience. Common patterns:

  • ENG team → audience Group: Engineering
  • OPS team → audience Whole tenant
  • Personal scratch team → audience Just me (use the Personal connectors flow)