Regulatory intelligence infrastructure

Regulations are a graph. Read them that way.

eCFR Cartography turns federal regulations into a live citation network that agents and analysts can traverse—edge by edge, version by version, grounded in the official eCFR API.

Explore Part 170 → Connect MCP

51Sections mapped

63Citation edges

2Structural hubs

1Stale candidate

01 / Live evidence

One part. Every relationship. Nothing hidden in the margins.

Interactive preview · drag to inspect

45 CFR Part 170 governs ONC health-IT certification. Node size reflects how often a section is referenced; links expose the part’s actual dependency structure. Search, drag, zoom, or select any section in the full workspace.

Connected Orphan Missing target

02 / Material finding

The graph catches what linear reading makes easy to miss.

Candidate stale reference § 170.599
→ 170.503

Part 170 still points to a section removed from the current regulation.

Section 170.599 incorporates a standard by reference for § 170.503. The target section was removed on June 30, 2020. Cartography surfaces the dangling edge, then lets an agent recover the prior text and amendment history. It is an investigation lead—not an automated legal conclusion.

03 / Agent surface

Six precise tools. One grounded model of the law.

01search_regulations

Search all 50 CFR titles and locate the part governing a topic.

02get_part_structure

Return the ordered structure of a part without downloading every section.

03get_section_text

Read one section, extract its citations, or recover its text at a prior date.

04map_part_references

Build nodes, edges, hubs, orphans, external citations, and statutory authorities.

05find_stale_references

Surface same-part citations whose target cannot be found in the current part.

06get_section_history

Trace every recorded amendment version for point-in-time investigation.

04 / Integration

From clone to first investigation in minutes.

Terminal · install

# Clone and create an isolated environment
git clone https://github.com/apraba05/ecfr-cartography-mcp
cd ecfr-cartography-mcp
python -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt
pytest -q
python server.py --list

Claude · MCP configuration

{
  "mcpServers": {
    "ecfr-cartography": {
      "command": "python",
      "args": [
        "/absolute/path/to/server.py"
      ]
    }
  }
}

05 / Investigation

A complete regulatory trace in five moves.

Discover

search_regulations
Locate Part 170 from a health-records API question.

Map

map_part_references
Resolve 51 sections into 63 explicit relationships.

Detect

find_stale_references
Surface the § 170.599 → § 170.503 candidate.

Prove

get_section_history
Confirm the target was removed on June 30, 2020.

Recover

get_section_text
Retrieve the pre-removal language at a chosen date.

Built for accountable AI

Evidence first. Always.

Authoritative source. Every answer starts with the official eCFR API, not a stale document corpus.

Reproducible extraction. Citation parsing is deterministic, offline-testable, and shared with the visualizer.

Honest uncertainty. Missing targets are candidates for review, never presented as legal conclusions.

Temporal context. Version history and point-in-time retrieval preserve what a rule meant when it applied.