The MCP integration is the structured bridge between Cooperly and external AI systems. It is designed for scoped access, explicit permissions, and predictable context delivery through Cooperly's /mcp endpoint rather than copy-pasted prompts or broad account exposure.
Overview
Cooperly's MCP layer exposes a remote Streamable HTTP endpoint for Codex and compatible clients. Each installation can be configured with human mode, service mode, or both, depending on whether the consumer is a person-driven assistant or a background workflow.
The model is intentionally explicit: installations, scopes, actor modes, and team grants are all configurable so access stays reviewable and bounded.
This is what turns Cooperly from a useful product UI into a reusable context layer for external AI workflows.
What context is available
The current MCP server exposes structured resources for team overview, current dashboard state, latest coop profile, latest fundamentals, team members, member summaries, full member profiles, and source-status freshness checks.
That means an assistant can work from live team context instead of static summaries. It also means the same installation can be useful for planning, drafting, onboarding support, and lightweight operations work without inventing a different integration for each.
How team connections work
Human-mode access is user-scoped and bounded by workspace surface enablement, Cooperly OAuth scopes, current product permissions, and team-level grants. A user can only read the teams Cooperly allows for that surface.
Service mode is aimed at background bots, scheduled workflows, and app-level automation. Service apps are owner-reviewed, and both modes still rely on explicit scopes and team-level grants, so the integration stays legible to operators and audit-friendly over time.
Configure
- 01
Enable MCP access in Cooperly
Make the developer MCP surface available for the account and intended teams.
- 02
Choose the clients and access mode
Decide whether the setup is for user-driven access, service access, or both.
- 03
Set scopes and team grants
Choose the Cooperly scopes and the team-level grants that installation should be allowed to use.
- 04
Connect the client and test context
Connect Codex or another compatible client, approve access, and confirm the expected team context is available.
Once an installation is live, Cooperly lets you reconnect, revoke, or rotate it without rebuilding the surrounding workflow from scratch.
That makes MCP a better long-term foundation for external AI access than one-off prompt plumbing or informal copy-paste context.