get_requirements returns normative requirements structured and filterable by level, which turns a spec from prose into a list an implementation can be held against; generate_checklist writes that list out as Markdown for a specific role and section set, and validate_statement runs a claim back through it. The metadata comes from the IETF Datatracker rather than the document text alone, so category, stream and publication date carry the authoritative record — with includeAuthors flagged as costing extra calls, which is a setting to make deliberately rather than by default.
An RFC reader with 7 tools that return a document's structure and metadata, extract its normative requirements and definitions, map its references, generate an implementation checklist, and validate a statement against it.
- get_rfc_structure returns the section hierarchy with metadata enriched from the IETF Datatracker — category, stream, publication date and abstract — and includeAuthors to resolve author names at the cost of extra API calls.
- get_requirements extracts normative requirements in structured form, filterable by level across MUST, SHOULD and MAY and by section.
- get_definitions returns term definitions from an RFC, optionally for one term.
- get_rfc_dependencies returns the reference graph, normative and informative, with includeReferencedBy for the inbound direction.
- get_related_sections returns the sections related to a given one.
- generate_checklist produces an implementation checklist in Markdown for a chosen role, over chosen sections, optionally including subsections.
- validate_statement checks whether a statement complies with an RFC's requirements.
Nothing — no account, no key. includeAuthors on get_rfc_structure costs extra Datatracker API calls.
One command — npx -y @shuji-bonji/rfcxml-mcp