§ MCP · Model Context Protocol

Give an agent the whole record

Every number and quote in this app comes from a live query — and the same data is exposed to AI agents over a Model Context Protocol server (insightbridge-aph-ai, 21 read-only tools). Add it as a connector and your agent can search every passage, split any topic by sector (who is speaking), find the places where sectors actually disagree, read what each submitter told the committee, and see who answered which of the committee's terms of reference.

Add it as a connector

Connector URLhttps://insightbridge-trainingdata.aicolab.org/mcp

Claude — personal accounts (Free, Pro and Max)

Go to Customize → Connectors — or open claude.ai/customize/connectors directly.
Click +, then Add custom connector.
Paste the URL above and click Add. No authentication is required — leave Advanced settings empty.
In a chat, click + (or type /), hover Connectors, and toggle it on for that conversation if it isn't already.

Connectors are account-level: added once, available in Claude on the web, desktop and mobile. Free accounts can add one custom connector.

Claude — organisation accounts (Team and Enterprise)

On Team and Enterprise plans only an organisation Owner (or Primary Owner) can add a custom connector — members do not see the option, so the personal steps above will dead-end.

An Owner opens Organization settings → Connectors — or claude.ai/admin-settings/connectors directly.
Click Add, hover Custom, then select Web.
Paste the URL above and click Add. No authentication is required — leave Advanced settings empty.
Each member then opens Customize → Connectors, finds the connector the Owner added, and clicks Connect; in a chat it is enabled via + → Connectors like any other tool.

ChatGPT — requires developer mode

Turn on Developer mode. OpenAI has moved this toggle more than once: look under Settings → Apps → Advanced settings, and if it is not there try Settings → Connectors → Advanced or Settings → Security and login. On Business, Enterprise and Edu workspaces an admin has to allow it under Permissions & Roles → Connected data.
Go to Settings → Connectors. The button to create a custom connector only appears once developer mode is on.
Add the URL above with transport Streamable HTTP and authentication None.

Available on Plus and Pro; on workspace plans an administrator can disable developer mode or allowlist specific connectors.

Microsoft 365 Copilot — via Copilot Studio

Microsoft 365 Copilot has no “add a connector” button for end users. The route is an agent built in Copilot Studio with this server as a tool, then published into Copilot. Anyone with a Copilot Studio licence (or trial) and an environment can build one.

Open copilotstudio.microsoft.com, create a new agent (or open one), and go to its Tools page.
Select Add a tool → New tool → Model Context Protocol.
Give it a Server name and a one-line Server description — the orchestrator reads the description to decide when to call it — paste the URL above as the Server URL, set authentication to None, and select Create.
On Add tool, choose Create a new connection, then Add to agent (or Add and configure to review the tools first — each can be switched off individually). If the first test run asks you to Open connection manager, select Connect then Submit and retry.
Publish the agent, then on the Channels page open the Teams and Microsoft 365 Copilot tile, keep Make agent available in Microsoft 365 Copilot selected, and select Add channel.
See agent in Teams → Add installs it for you. In Microsoft 365 Copilot, type @, pick the agent, and ask. For the whole organisation, use Availability options → Show to everyone in my org → Submit for admin approval.

Copilot Studio speaks Streamable HTTP only (SSE was dropped in August 2025), which is exactly what this endpoint offers. Behind the wizard it creates a Power Platform custom connector, so tenant data-loss-prevention policies on connectors apply and an administrator may need to allow it. Tenants that want central governance can instead have an admin register the URL with Agent 365 (a365 develop-mcp register-external-mcp-server, auth type NoAuth) and approve it under Agents → Tools → Requests in the Microsoft 365 admin centre; it then appears in every Copilot Studio environment as a registry entry. That path is in preview.

Claude Code

One command — --transport http is what makes it a remote connector rather than a local process:

claude mcp add --transport http training-data https://insightbridge-trainingdata.aicolab.org/mcp

Re-adding after a change needs claude mcp remove training-data first.

Then ask something like “Using the AI inquiry record, where do companies and civil society submitters disagree about copyright and training data, and who said what?”

Works with any MCP client. The endpoint speaks JSON-RPC over a stateless one-POST Streamable-HTTP adapter at POST /mcp; there is no SSE stream and no authentication, so a client that insists on OAuth should be set to “no auth”.

How to read what comes back

Submitter counts are not prevalence. These are the submitters who chose to write to a parliamentary inquiry. "N submitters say X" measures the record, never a population.
Positions are relative to each topic's own proposition — Supports, Builds on, Mixed, Redirects, Opposes, Unclear measure stance-toward-a-framing. Most read supportive, so agreement is the default and only contested_topics finds real divergence.
One lens: sector. Company · Industry Group · Individual · Civil Society · Government · Academia. There is no time axis; every document dates from the 2026 inquiry.
One grouping generation. Seven themes sit straight on the topics, so list_families and list_superclusters return the same nodes. The tools stay so the same connector works against a re-run with a deeper tree.
The terms of reference are answered per document. An unanswered clause means no answer was located in the available text, never that a submitter has no view; merely quoting the terms does not count.
One source is partly unreadable. Submission 146 has black pages in the published original; its document carries extraction_quality = limited_by_source and its text is partial.
topic_id is the topicClusterId, not the display cluster_id; both look plausible, so the wrong one quietly returns a different topic.

Try the tools

Pick a tool, adjust the inputs, and run it against the live server — the exact request an agent would make. Every tool is pre-filled with example values that return real results.

Orientation & search

The grouping tree

Anatomy of a topic

Divergence & silence

Sectors & submitters

Documents

The terms of reference

Facets

corpus_overview

No inputs — just run it.

POST /mcp

{
  "method": "tools/call",
  "params": {
    "name": "corpus_overview",
    "arguments": {}
  }
}

Response

Run a tool to see its output.