Model Context Protocol

UK procurement intelligence for AI agents

Public notices give you the facts. TenderLedger gives Claude, Cursor, and enterprise copilots the next move — without raw database access or the full product graph.

Controlled tools for market summaries, buyer patterns, renewal signals, and competitor comparisons.

Request API access Browse tools

What the MCP server exposes

The Model Context Protocol gateway exposes controlled intelligence tools backed by official UK sources — Contracts Finder, Find a Tender, and devolved portals. Agents get structured, capped responses designed for reasoning — not scraping.

Market summaries with top buyers and suppliers
Renewal and re-tender signal teasers
Pursuit qualification when a tender id is known
Market change slices over configurable windows

agent → tenderledger-mcp

{
  "tool": "search_procurement_market",
  "arguments": {
    "query": "NHS cyber security",
    "limit": 5
  }
}

→ market_summary
→ top_buyers[5]
→ recent_awards[5]
→ renewal_count
→ tenderledger_url

Example tools

Seven focused tools — each returns structured JSON with attribution and a CTA back to TenderLedger.

search_procurement_market

High-level market slice for a topic

e.g. NHS cyber security

get_buyer_intelligence

Compact buyer profile from indexed notices

e.g. Ministry of Defence

get_supplier_intelligence

Competitor profile from disclosed winners

e.g. Softcat

find_likely_renewals

Re-tender teasers from precomputed signals

e.g. facilities management · 90 days

qualify_opportunity

Pursuit / bid qualification score

e.g. tender id or topic slice

get_market_changes

What moved in the last N days

e.g. IT services · 14 days

compare_suppliers

Side-by-side comparison for up to three names

e.g. Softcat vs Capita vs Atos

Example agent prompts

Natural-language questions your copilot can route to TenderLedger tools.

“Which suppliers are winning MOD-linked IT awards in the TenderLedger index?”
“Summarise fire safety renewal signals for the next 12 months.”
“Compare Softcat and Capita on disclosed public-sector awards.”
“What changed in NHS cyber security procurement in the last two weeks?”

Safety limits

Capped responses

Typically five items per dimension — no raw table dumps or unlimited exports.

Confidence-labelled

Every payload includes source, freshness timestamp, and a link back to TenderLedger.

Quota tiers

Anonymous IP limits, API keys for higher daily quotas, Pro keys for team workloads.

No private data

No arbitrary SQL, no access to other users' workspaces or full intelligence graph.

How to run locally

Full setup docs live in backend/MCP.md. Quick start for Cursor / Claude Desktop (stdio) or REST invoke:

Cursor MCP config

{
  "mcpServers": {
    "tenderledger": {
      "command": "python",
      "args": ["-m", "app.mcp.stdio"],
      "cwd": "/path/to/backend"
    }
  }
}

REST invoke

curl -X POST /api/mcp/v1/invoke \
  -H "X-TenderLedger-MCP-Key: your-key" \
  -d '{"tool":"search_procurement_market",
       "arguments":{"query":"NHS cyber"}}'

1Enable MCP on the API host (TENDERLEDGER_MCP_ENABLED=true)
2Configure stdio for Claude Desktop / Cursor, or use REST invoke
3Authenticate with X-TenderLedger-MCP-Key for production quotas
4Call tools from your agent — responses loop back to TenderLedger for deeper workflows

Works with

Claude DesktopCursorCustom copilotsEnterprise agents

Give your agents procurement intelligence

MCP is gated by configuration on the API host. Request production keys, higher quotas, or enterprise audit requirements.

Book a demo Start free trial

Slack integration · Pricing · Intelligence hub

Stop browsing notices manually.

Start prioritising the contracts you can actually win.

Start Free Trial

Built on official UK procurement sources