Bankruptcy docket MCP server

#What the server is

ElevenFlo MCP is a hosted Model Context Protocol server for source-backed chapter 11 research: research tools plus monitor management. It connects Claude, ChatGPT, and other MCP-compatible clients to case records, dockets, filing text, hearing transcripts where indexed, document summaries, document relationships, and published restructuring datasets. The connection uses OAuth. It does not use pasted API keys.

The underlying corpus includes 2.7M+ live court documents. A live document is a filing that ElevenFlo still holds. It does not mean that every page has searchable text or that every case has complete coverage.

Start with the setup guide to connect your client. ElevenFlo MCP returns evidence and identifiers to your client. The client decides how to present the result, preserve a conversation, or run a prompt.

#Public research tools

Use exact public tool names because the server accepts no aliases. The basic research loop is:

• Call find_cases to resolve the correct case_id.
• Use list_docket_entries, search_filings, or search_transcripts to locate source material.
• Use search_document_summaries and get_document_summary for orientation.
• Use search_document_chunks, read_document_chunks, or extract_document_passages for supporting filing text.
• Use find_key_documents and get_related_documents to trace connected motions, objections, hearing materials, and orders.

Taxonomy and structured data use list_document_types, list_structured_datasets, describe_structured_dataset, search_structured_data, aggregate_structured_data, and suggest_structured_values.

Monitor management uses save_monitor, list_monitors, and delete_monitor on the signed-in user's own monitors: a watched case, or market filters such as chapter, state, status, and asset class. Call list_case_updates to read what changed for them, including corrections and removals, and keep its next_cursor to resume later.

#Public datasets

The public server exposes structured datasets covering selected records such as voluntary petitions, committee appointments, sale processes, claim transfers, bar dates, exclusivity periods, case outcomes, and plan voting. Call list_structured_datasets for the available slugs, then call describe_structured_dataset before a search or aggregation. The description defines the public fields, filters, sorts, facets, and supported operations.

Structured rows answer typed-field and comparison questions. They do not replace the filed document. A result includes freshness, coverage, and provenance fields so your client can describe what was queried and what may be missing.

#Verification boundary

Treat docket metadata, summaries, and retrieved text as different evidence levels. For legal language, dates, amounts, deadlines, releases, injunctions, or defined terms, retrieve the relevant passage and cite the filing. A missing row or search result does not prove that an event did not occur.

Apart from the user's own monitors, the server changes nothing. It cannot file documents, modify a docket, send email, change account settings, or provide legal or financial advice. For a complete worked example, use the daily case brief.