ietf-llm

Maintain a local, queryable corpus of an IETF effort's public record — a Working Group, an IRTF Research Group, a mailing list, or a set of drafts. It pulls together charter, drafts, RFCs, meeting agendas, minutes, slides, transcripts, attendance, mailing list archives, and GitHub issues and pull requests for use with LLM-based tools.

What it's for

A working group's history is spread across mailing list archives, Datatracker, GitHub, and meeting materials — too much to hold in your head, and too scattered to search well by hand. With the record gathered into one queryable corpus, an LLM can help you:

• Get up to date with the state of discussions — what's open, what was recently decided, where a debate currently stands.
• Summarise the arguments already made about an issue — every distinct position on a topic, who holds it, and how the chairs ruled.
• Formulate a new proposal — surface the objections raised against similar ideas before, so you can anticipate them.
• Fact-check assertions about what's happened so far — grounded in the actual list traffic and chair statements, not someone's recollection.

Quick start

1. Install with pipx:

   pipx install 'ietf-llm[local-embeddings]'
   

2. Gather a group (here httpbis; an IRTF Research Group such as cfrg works the same way):

   ietf-llm httpbis
   

3. Register the MCP server with your client — for Claude Code:

   claude mcp add ietf-llm -- ietf-llm-mcp
   

   Other clients are covered in Running the MCP server locally.

4. Ask your client about the group.

What you can ask

The norms tools need nothing gathered. The RFC and chartered-group tools need the series data that ietf-llm --init or any gather installs; the rest read a corpus you've gathered. draft_status, review_record and meeting_schedule also look up current state on the Datatracker. The full tool list is served to the client with the server's instructions.

• Orientation — what's gathered, what a group is doing, who's active, what's open.
• Chartered groups and RFCs — which group or RFC covers a topic; charters, milestones, history, and RFC text across the whole series.
• Discussion — what was argued about a topic, replies to a message, chair calls and polls, and whether something was ever said or proposed.
• Issues, pull requests and drafts — filtered catalogues, verbatim draft text by section, and why the text says what it does.
• Meetings — minutes with poll tallies, attendance, the upcoming schedule.
• Norms — how to read the record without overclaiming consensus, and how to contribute without wasting the group's time.

Ways to run it

• Local MCP server — one server per client, on your own machine. Best starting point.
• Shared HTTP MCP server — one hosted process serving many clients.
• CLI search — semantic search over the cache with ietf-llm-search, no LLM client needed.
• NotebookLM export — mirror a corpus to files, or push it to NotebookLM Enterprise.

Every mode reads a corpus you first gather with ietf-llm <corpus>. The full documentation listing is in docs/README.md.

Gathering reaches the network. To make a first gather fast, ietf-llm pulls a prebuilt snapshot of covered corpora from the public seed store at seed-store.mnot.net, then freshens only the delta; list_corpora looks its catalog up there too. To turn this off, run ietf-llm <corpus> --no-seed (it persists across gathers; --seed re-enables it), or set IETF_LLM_SEED_ENABLED=off or IETF_LLM_SEED_URL=off. With seeding off, gathers run fully cold. An unreachable or unlisted corpus just gathers from scratch.