Search GovTribe

Search_GovTribe is the integrated GovTribe finder tool for broad cross-dataset discovery when the caller needs one ranked result list first and can resolve details with typed Search_* tools afterward.

When To Use

Typical questions this tool answers well:

• Which GovTribe records across contracts, vendors, grants, news, files, and workspace entities best match this query?

• What is the best first-pass finder when the user has not chosen a specific GovTribe dataset yet?

• Which candidate records should be resolved next with typed Search_* tools for richer fields, filters, and pagination?

• Which mixed public and workspace records are most relevant to a conceptual or exploratory query?

Required Reading

1. Search_Mode_And_Query_Guide: Required before setting free-text query strings because this tool uses the caller query directly and does not rewrite it with another LLM call.

Output Contract

• Top-level keys:

  • query: The exact input query string.

  • retrieval_strategy: The integrated ranking pipeline used for this response. Current values are hybrid_rrf_then_ai_collection_rerank on successful rerank and hybrid_rrf_with_keyword_fallback when reranking is skipped or falls back.

  • matched_types: GovTribe types that contributed at least one result row.

  • results: Ranked finder rows.

• Row keys:

  • rank

  • score

  • matched_retrievers

  • govtribe_id

  • govtribe_type

  • title

  • url

  • snippet

  • resolver

• matched_retrievers values:

  • keyword

  • semantic

• score meaning:

  • this is the integrated base ranking score used by GovTribe’s internal retrieval pipeline

  • the final AI rerank stage reorders rows but does not expose provider rerank scores in this tool output

• resolver keys:

  • tool

  • filter

  • arguments

  • suggested_fields

  • documentation_articles

• Resolver-followup behavior:

  • use resolver.tool as the next typed GovTribe retrieval tool

  • pass resolver.arguments into that tool directly

  • request richer fields with resolver.suggested_fields

  • consult resolver.documentation_articles before building more complex typed-tool calls

Usage Patterns

Pattern A: Broad integrated discovery across all eligible datasets. Tool: Search_GovTribe

{
  "query": "\"marine boatyard services\""
}

Pattern B: Narrow the finder to a mixed but relevant type subset. Tool: Search_GovTribe

{
  "query": "companies and awards related to ship repair and dry dock support",
  "max_num_results": 8,
    "govtribe_types": [
      "vendor",
      "federal_contract_award",
      "state_local_contract_award",
      "government_related_news_article"
    ]
  }

Pattern C: Use the integrated finder as the first half of a find-and-resolve workflow. Step 1 Tool: Search_GovTribe

{
  "query": "\"orion\" marine repair",
  "max_num_results": 5
}

If a result returns: Step 2 Tool: Search_GovTribe

{
  "govtribe_id": "vendor_123",
  "govtribe_type": "vendor",
  "resolver": {
    "tool": "Search_Vendors",
    "filter": "vendor_ids",
    "arguments": {
      "vendor_ids": ["vendor_123"]
    }
  }
}

Resolve it with the typed tool: Step 3 Tool: Search_Vendors

{
  "vendor_ids": ["vendor_123"],
  "fields_to_return": ["govtribe_id", "name", "uei", "govtribe_url", "govtribe_ai_summary"]
}