The chain that makes this worth wiring in is search, rank, extract requirements, then export a shortlist and a calendar file. Requirement extraction fetches the specification document itself, which is the step that normally means downloading a PDF and reading it by hand.
A read-only server over Japan's public procurement notice search, plus scoring, requirement extraction and export on top. Free tools cover search and ranking; the rest are on the paid tier.
- search_bids, list_recent_bids and get_bid_detail — keyword, region and category search over notices
- rank_bids — score candidates against your own criteria
- explain_bid_fit and assess_bid_qualification — why a notice fits, and whether you meet the entry requirements
- extract_bid_requirements — requirements pulled from the specification PDF or HTML
- analyze_past_awards and summarize_bids_by_org — award history and a per-agency view
- create_bid_review_packet, draft_bid_questions, export_bid_shortlist and create_bid_calendar
- save_search, check_saved_search and list_saved_searches for standing queries
- search_bids_app, which returns an interactive panel in clients that support it
- Resources and resource templates addressing a bid, a prefecture or an organization directly
Local use needs nothing but npx and the package jp-bids-mcp over stdio, with no authentication. The remote endpoint at https://mcp.bid-jp.com/mcp requires OAuth 2.0 — clients that run the flow handle it for you, and a plain request without a token gets a 401 pointing at the metadata. Data comes from the Small and Medium Enterprise Agency's procurement portal.
One command — npx --yes jp-bids-mcp
