Back to Guides

/gcloud-kit:tenders

Generate per-service UK procurement intelligence from live awarded-contract data.

Overview

Queries the UK Find a Tender Service (FTS) and Contracts Finder for awarded contracts related to your service's category. Produces a tender intelligence report covering awarded-value benchmarks, incumbent supplier patterns, market concentration, and comparable awards with notice URLs. Use this to validate pricing, identify who actually wins work in your space, and understand how buyers procure your category.

Usage

/gcloud-kit:tenders 001
/gcloud-kit:tenders payroll

Arguments: Service number or name fragment (must match an existing service directory).

Prerequisites

  • Service design (services/{NNN}-{name}/service-design.md) — required; CPV codes and category are derived from it
  • Bundled uk-tenders-mcp MCP server — deferred, works out of the box; the command degrades gracefully if the live endpoint is unavailable

Output

services/{NNN}-{name}/tender-intel.md

The report contains:

  • Awarded-value benchmarks — median, lower quartile (Q1), upper quartile (Q3), and sample size for comparable GBP awards
  • Incumbency & concentration — the top incumbent and its share of awarded value, plus a concentration band (LOW / MEDIUM / HIGH)
  • Comparable awards — a table of individual awarded notices: award date, buyer, supplier, value, and a link to the official notice URL
  • Top suppliers in this space — suppliers ranked by award count and total awarded value
  • Top buyers in this space — buyers ranked by award count and total awarded value
  • Gaps & caveats — CPV codes with no awards found, MCP errors, or notices that could not be retrieved

If the uk-tenders-mcp endpoint is unavailable, the command still produces the artefact — the award tables will be empty and the reason recorded under Gaps & caveats.

What it does

  1. Reads the service design to extract category, CPV codes, and lot context
  2. Queries UK tender data (FTS + Contracts Finder) via the uk-tenders-mcp tools
  3. Filters and ranks results by relevance to the service's CPV mapping
  4. Calculates value benchmarks and identifies concentration patterns
  5. Assembles the tender intelligence report and writes it to the service directory

Architecture: three-tier subagent split

The command uses three tiers to keep concerns cleanly separated:

TierRoleBoundary
OrchestratorReads service design, drives workflow, reviews outputNo MCP tools, no file writes
Tenders-readerSole caller of uk-tenders-mcp tools; returns structured dataUntrusted MCP input boundary — isolated here
Tenders-writerSole user of the Write tool; formats and persists the reportNo MCP calls

The reader isolation matters because MCP responses are untrusted external input. Confining all uk-tenders-mcp calls to the reader tier prevents prompt-injection or malformed data from reaching the orchestrator or triggering unintended file writes.

Caveats

  • Awarded value ≠ actual spend — contract values are the notified ceiling at award time; actual spend may differ
  • CPV ↔ lot mapping is approximate — CPV codes do not map exactly to G-Cloud lots; some awards may be tangential
  • UK public-sector only — data covers UK contracting authorities; private-sector procurement is not included
  • OGL v3.0 attribution — tender data is published under the Open Government Licence v3.0; downstream use must retain attribution

Related commands

  • /gcloud-kit:compare — shows who is listed on the Digital Marketplace; tenders shows who actually won public contracts — use both for a complete competitive picture
  • /gcloud-kit:pricing — use awarded-value benchmarks from this report as a sanity-check when setting day rates or contract values
  • /gcloud-kit:service-design — creates the service design that this command reads for CPV and category context
  • /gcloud-kit:sales — spend data from G-Cloud itself; complements tender data from wider procurement routes