Connect Your Tools

Modules 02 and 03 gave you a Team Brain and routines that refresh it. Both depend on the AI being able to read where work happens — Jira, Notion, Drive, Slack, your repo, your team-brain folder. Module 04 wires those connectors, with scope-narrowing patterns that keep the AI from accidentally surfacing private channels, NDA documents, or that one Slack DM nobody wants in a Friday digest.

Three artifacts — connectors.md, scope-rules.md, audit-policy.md — that document what's wired, what's narrowed, and what's logged. Runs three ways:

• Platform connectors — your AI tool's built-in connector marketplace. Best for: Claude Connectors (MCP-based), ChatGPT Apps (Drive · Slack · GitHub · Atlassian Rovo), Gemini Workspace, M365 Copilot.
• Self-hosted MCP server — you run your own MCP that exposes internal tools with custom scopes. For teams with on-prem systems or a "no third-party connector" policy.
• Manual scoped pastes — for sensitive data: human-curated paste-from-Jira / paste-from-Slack at session start. No persistent connector. The most conservative option.

Read-only by default. Write access is per-routine, per-purpose, with audit. The point is to give the AI the same access a human has — not more.