TenderMetro for Agents

For agents using TenderMetro MCP tools to support tender research, capture and market-intelligence work. TenderMetro is built for agents, not only for people.

# Add TenderMetro as a remote MCP server
# Claude, ChatGPT, Microsoft Copilot, Cursor or any MCP client
https://app.tendermetro.com/mcp

Welcome

Built for agent-assisted capture work

TenderMetro gives agents structured tools, record context and response references so they can help users research markets, qualify opportunities, organise capture work and explain source data without guessing.

Tool pages

Define the available actions, valid arguments and task-shaped examples for TenderMetro AI and the TenderMetro MCP server.

Response contracts

Describe payload fields, status objects, relationships and interactive outputs that agents can cite or pass to follow-on tools.

Guides

Explain shared patterns for search construction, filters, context control, saved searches and grounded retrieval.

Data model links

Connect tool output back to TenderMetro record types, fields, identifiers and lifecycle details.

Guides

Cross-tool guidance for search construction, filters, context control, saved searches and grounded retrieval.

Jump to section

MCP tools

Callable tools that search TenderMetro data, manage workspace records and return market intelligence.

Jump to section

MCP tool responses

Record payloads, status snapshots and output contracts agents can cite, inspect or pass to follow-on tools.

Jump to section

Connect

The MCP server

One remote endpoint carries every TenderMetro tool. Users connect their own account, so agents act with the user's permissions and plan limits.

Connect

1. Add https://app.tendermetro.com/mcp as a remote MCP server in the client.

2. The user signs in to TenderMetro and approves the connection on the consent screen.

3. Tools become available in the conversation; workspace actions run as that user.

Good to know

• OAuth 2.1 consent — no API keys to paste into chat clients.
• Read tools are safe to call freely; write tools ask for confirmation.
• Tool output mirrors the REST API and the data model — same fields, same enums.
• Rate limits follow the user's plan and are reported in every error response.

Guides

Agent guides

Cross-tool patterns for building TenderMetro MCP requests: search construction, filtering, retrieval, saved searches and grounded answers.

Search construction

Translate a plain-language research question into a filtered search: keywords first, then sector, region, country, status and value to narrow.

Filtering & context control

Keep result sets small enough to reason over — ask for limits, page through offsets, and pull full records only for shortlisted references.

Saved searches

Reuse the user's saved searches as standing queries instead of rebuilding filters, and read their alert schedules for monitoring intent.

Grounded answers

Cite the source and reference on every record you surface. Never invent deadlines, values or buyer names — fetch the record first.

Capture actions

Confirm with the user before writing: tracking an opportunity or updating a pursuit stage changes their workspace.

Aggregations & stats

Use market_stats for totals and breakdowns by sector, region or country instead of summing raw rows yourself.

Reference

MCP tools

The callable tools exposed by the TenderMetro MCP server, with the arguments each accepts.

Tool	Arguments	What it does
search_opportunities	query, sector, region, country, status, min_value, limit, offset	Keyword and filter search across every tracked market.
get_opportunity	reference	Full tender record including description, buyer and documents.
get_buyer	slug	Buyer profile with recent tenders and award history.
get_company	slug	Supplier profile with award history and total awarded value.
market_activity	kind, since, limit	Latest market signals: notices published, closing soon, awards.
market_stats	group_by (sector \| region \| country)	Totals and awarded value across the index.
my_pipeline	stage, limit	The signed-in user's pursuits with stages and values.
my_saved_searches	—	The signed-in user's saved searches and alert schedules.
track_opportunity	reference, stage, notes	Add or update a tender in the user's pipeline (write — confirm first).

Responses

MCP tool responses

What tool output looks like and how agents should use it.

Record payloads

Tools return the normalised record types from the data model — opportunity, award, buyer, company, document — with the same fields shown in the app and the REST API.

Status objects

Every response includes status, deadline and value semantics in UTC, plus the quality score so an agent can judge completeness before relying on a field.

Relationships

Payloads carry references (buyer slug, opportunity reference) that are valid arguments to follow-on tools, so agents can walk the graph without free-text lookups.

Citations

Each record includes its source system and original reference — agents should surface these so users can verify against the publisher's own site.

Record types, fields and identifiers are defined in the data model — tool output connects directly back to those definitions. REST endpoints and webhook contracts are in the developer docs.

Put an agent on your pipeline

Connect ChatGPT, Claude, Copilot or your own agents to live global tender intelligence.

MCP server details Developer docs