Connect Moral Fabric to your AI tools (MCP + AI context)
Connect Moral Fabric to your AI tools (MCP + AI context)
Moral Fabric can plug into any AI assistant that supports the Model Context Protocol (MCP) — Claude, ChatGPT (via custom GPTs / connectors), Cursor, and a growing list of others. We use Claude internally, so it's our reference example throughout this guide, but the connector itself is tool-agnostic.
Once connected, your AI assistant can answer questions about your workspace — teams, roles, people, projects, suppliers, role-fit ratings, and 360 feedback — without you switching tabs or copy-pasting. It's read-only and respects your permissions — same rules as in the app, see Who can see your data.
Option 1 — Add the MCP connector (recommended)
This is the live, two-way option: your AI assistant queries fresh data on demand.
Connector URL:https://mcp.moralfabric.org
In Claude (Teams / Enterprise)
Settings → Connectors → Add custom connector
Choose Web (Remote MCP)
Paste the URL above
Click Connect → log in to Moral Fabric → approve
In other tools
ChatGPT (Enterprise/Team): Settings → Connectors → Add MCP server
Cursor / Windsurf / Zed: add to your mcp.json config
Anything else MCP-compatible: point it at the URL, follow its OAuth flow
If you belong to multiple Moral Fabric workspaces, just mention which one in your prompt (e.g. "In the SMA workspace, who leads the Fundraising team?").
What you can ask
Quick lookups
"Show me the full org structure."
"What roles does Anna hold, and which teams is she in?"
"Which suppliers fill the Bookkeeping role?"
Combining data types inside Moral Fabric
This is where it gets interesting — your AI assistant can pull from multiple parts of your workspace in one go:
"Who rates their energy as 'Drains' on multiple roles? They're probably overloaded." (uses role ratings)
"For everyone in the People team, show their role-fit ratings alongside their most recent 360 feedback. Where do they line up, where do they conflict?" (combines ratings + 360 feedback)
"Which teams have projects but no assigned role? Those are orphans we should clean up."
"Compare my role portfolio to Anna's — overlapping accountabilities, blind spots, possible coaching matches."
"List every supplier filling more than three roles. Where are we vendor-concentrated?" (uses supplier links)
"For the Fundraising team: roles + assignees + suppliers + open projects, on one page."
Combining Moral Fabric with other MCPs
If you've also connected Asana, Slack, Google, Notion, etc. in the same AI assistant, it can stitch them together — Moral Fabric becomes the "who and what" backbone:
"Look up the Asana project 'Spring campaign' and tell me which Moral Fabric role is responsible, who that is, and what their energy rating on that role is." (works best when Asana is also synced into MF)
"Find a 45-minute slot next week (Google Calendar) with every Lead Link of every team in Moral Fabric."
"Draft a Slack DM to all assignees of the 'Bookkeeping' role across both my workspaces — keep it casual."
"For each role in the Tech & Tools team, search Google Drive for related policy docs and tell me what's missing."
"Cross-check: which suppliers in Moral Fabric appear in our Moneybird invoices but aren't yet linked to a role?"
"From Notion, pull the onboarding doc, then list every Moral Fabric role it mentions — and flag the ones that have no current assignee."
Strategic / reflective
"Based on role accountabilities and recent 360 feedback, where might our org be under-staffed or duplicating effort?"
"Looking at all coaching pairs and their coachees' role ratings, who would benefit from a different coach match?"
Option 2 — Paste the AI Context URLs
If your AI tool doesn't support MCP (e.g. a plain Claude Project, a Custom GPT, an internal LLM setup), you can give it our two reference URLs instead. They're a static snapshot — no live querying — but they make the assistant fluent in how Moral Fabric is structured.
Where to find them: in the app, go to Settings → Advanced → AI Settings → AI Context for Tools. You'll see two copy-able URLs:
URL
What it contains
App context
What Moral Fabric is — concepts like "circles", "pattern roles", "adoptions", "role ratings". Safe to share with any AI; contains no workspace data.
Workspace context
Your specific circles, roles, and people. Treat this one like internal data — same access boundary as in Who can see your data.
Paste either (or both) into the system prompt / project knowledge of your AI tool, and it will understand the platform's vocabulary and your org's structure when you ask questions.
If you don't see the URLs there, your platform admin needs to set the AI_CONTEXT_TOKEN environment variable. Email us — see below.
Troubleshooting
"Not connected" or login loop → disconnect the connector and re-add it; tokens occasionally expire.
"No workspace specified" → you belong to more than one workspace. Mention the workspace name in your prompt.
Tool calls hang → make sure you added the connector as Remote / Web MCP, not stdio.
My tool isn't on the list above → the MCP standard is universal, so most likely it works. Check your tool's docs for "MCP server" or "remote tool" support.