Smart Tender Procurement Search

Problem

Tender search often fails because keyword-only matching mixes maintenance notices, consumables, and stale announcements, and makes it hard to distinguish who issued the bid, who won, and who participated. This skill connects nationwide tender, procurement, and winning-bid data into the workflow, so engineers can locate projects by region, amount, time, industry, and company role, then continue into company profiling, competitor mapping, and brand or price trend analysis.

How It Works

It calls a POST-based backend API with several practical capabilities:
- Basic search: search_bids filters by keyword, region, amount, time, and other dimensions; get_bid_detail retrieves the full announcement text.
- Structured logic: query_bids_advanced supports keywords, keyword_groups, and exclude_keywords, for example requiring “server” while excluding “maintenance.”
- Role matching: match_modes separates caller, winner, tender, and fulltext, preventing confusion between procurer, supplier, and bidder.
- Analysis chains: It can extend from one company to partners, competitors, top purchasers, top suppliers, top brands, and price trends.

Boundaries

The skill is best for structured tender intelligence. Result quality depends heavily on keywords and match modes; open-ended questions such as market trends or policy impact should be cross-checked with external sources. Calls require a valid ZLBX_API_KEY, and quota or rate-limit errors should be handled according to the returned status rather than by blind retries.