Data2Bids MCP Connector

What Claude can do once you connect your Data2Bids account.

What it does

Data2Bids ingests construction bid and project documents — drawings, specifications, engineering PDFs, and related files — and processes them into a searchable index and a knowledge graph. The MCP connector lets Claude query that index and graph on your behalf, so you can ask questions about your projects in plain language instead of opening files manually.

Read-only by design

Every tool exposed by this connector is a lookup: it lists, searches, resolves, or retrieves information that already exists in your account. None of the 18 tools create, modify, or delete data. All 18 tools are annotated readOnlyHint: true in the MCP tool schema.

Authentication

Access is granted through OAuth 2.0. When you connect, you authorize Claude to act as your Data2Bids user; the connector only ever sees the companies and projects your account already has access to. You can revoke the connection at any time from your Data2Bids profile.

Server

https://www.data2bids.com/mcp

Tools

Project & investigation

• data2bids_investigate — The entry point for an ambiguous question. Takes the user's literal message, tolerates typos, resolves which project it refers to within the authenticated company, and runs the graph investigation in the same call when confidence is high. If several projects could match, it returns candidates so Claude can ask which one the user means.
• projects_list — Lists the projects the authenticated user can access, with filters for status, visibility, and search text.
• projects_get_status — Returns the operational/processing status of a project and whether it's ready to be queried.

Files

• files_list — Lists the indexed files in a project.
• files_get_content — Returns the canonical extracted text content of a file.
• files_get_content_v2 — Returns enriched file content, including multimodal artifacts when they exist.
• files_get_content_page — Returns a paginated window of a file's content, for reading large documents incrementally or jumping to a specific section.

Semantic retrieval

• retrieval_search_by_embeddings — Semantic search across every file in a project for passages relevant to a query.
• retrieval_get_matches_for_file — Semantic search scoped to a single file.

Knowledge graph (GraphRAG)

These tools work over a knowledge graph built from a project's documents — entities such as bids, sheets, spec sections, change directives, requirements, and the relationships between them.

• graphrag_projects_list — Lists accessible projects and indicates which ones have a graph built.
• graphrag_files_list — Lists a project's files that have graph artifacts.
• graphrag_project_context_investigate — Full-project investigation: returns the relevant entities, relationships, file coverage, ranked passages, conflicts, and revision history for a question.
• graphrag_graph_resolve — Resolves a name or identifier to graph entities, returning candidates without merging similar-looking identifiers.
• graphrag_graph_search — Searches graph entities by text and type across a project.
• graphrag_graph_expand — Expands an entity's relationships by one hop.
• graphrag_graph_path — Finds an evidence-backed path of up to three hops between two entities.
• graphrag_graph_timeline — Returns revision events (modifications/supersessions) for an entity, so a later revision is only treated as authoritative when there's explicit evidence for it.
• graphrag_graph_evidence — Returns the underlying evidence (file, page, region, text, method, confidence) backing an entity or relationship.

Data handling

See our Privacy Policy for how document data is stored, processed, and retained.

Support

Questions about this connector: support@data2bids.com