splitifi-mcp
v2.4.4
Published
Splitifi Intelligence MCP — outcome predictions, judge profiles, and legal workflows powered by 29,211 production ML models trained on real court records. Serves litigants, attorneys, judges, mediators, CDFAs, and litigation funders.
Maintainers
Readme
Splitifi Intelligence MCP
Connect Claude to Splitifi's legal AI engine — 29,211 production ML models (37,351 total registry entries) trained on real court records, 16,302 verified judges, and outcome data from 2.5M+ analyzed cases.
Built for attorneys, litigants, litigation funders, CDFAs, mediators, and judges navigating family law, commercial litigation, and beyond.
What You Get
Tools (190+)
- Outcome predictions: custody, asset division, spousal support, child support, 30+ legal verticals
- Judge intelligence: grant rates, reversal rates, behavioral patterns, motion timing
- Comparable case analysis: outcome distributions across 2.5M+ real cases
- Calculators: child support (all 50 states), spousal support, divorce cost, home buyout, co-parenting cost
- Strategy paths: negotiation vs. litigation analysis, BATNA, settlement windows
- Financial forensics: EDGAR defendant screening, hidden asset flags, income suppression detection
- International screening: UK, Canada, Australia, EU corporate, UN sanctions, UAE, Israel, MENA
- Forms and checklists: jurisdiction-specific court forms, guided document checklists
- Litigation finance: IRR modeling, award range prediction, recovery probability
Prompts (power-user workflows)
analyze_case_senior_partner— full IRAC analysis with Splitifi predictions woven indraft_settlement_brief— structured settlement memo with BATNA and walk-away thresholdprepare_motion_filing— pre-file judge + precedent + motion score memolitigant_triage— 5-task next-step plan for pro se litigants
Resources (reference data)
splitifi://stats/platform— authoritative model and data countssplitifi://model/{name}/card— training data, ECE, AUC for any production modelsplitifi://jurisdiction/{code}/guide— divorce law reference by state or country
Setup
1. Get an API key
Create an account at mysplitifi.com/developers. Keys start with sk_live_ (production) or sk_test_ (sandbox — all tools unlocked, synthetic responses).
2. Add to Claude Desktop
Open claude_desktop_config.json (Mac: ~/Library/Application Support/Claude/, Windows: %APPDATA%\Claude\):
{
"mcpServers": {
"splitifi": {
"command": "npx",
"args": ["-y", "splitifi-mcp"],
"env": {
"SPLITIFI_API_KEY": "sk_live_your_key_here"
}
}
}
}Restart Claude Desktop. You should see the Splitifi tools appear in the tool selector.
3. (Optional) Point at a local backend
"env": {
"SPLITIFI_API_KEY": "sk_test_local",
"SPLITIFI_API_URL": "http://localhost:8080/api/v1/intelligence"
}Tool Scoping
At startup, the server calls your account's /me endpoint to fetch enabled_verticals. Only tools within your plan's enabled verticals are registered. Sandbox keys (sk_test_*) unlock all verticals.
If the backend is unreachable at startup, all tools are registered and the backend enforces authorization at call time.
Example Prompts
Custody prediction:
"Using Splitifi, predict the likely custody outcome for my client in Hillsborough County, FL. 12-year marriage, 2 children ages 6 and 9, father is primary earner ($140K), mother was primary caregiver, no DV history."
Judge intel before a motion:
"Use the Splitifi prepare_motion_filing prompt for case ABC-123, motion_type: temporary_support"
Settlement analysis:
"Analyze this settlement offer using Splitifi: $280K lump sum, no alimony, 60/40 custody. Case is in Palm Beach County, FL, 8-year marriage, combined assets $520K."
Litigation funder due diligence:
"Run Splitifi EDGAR screening on Acme Corp as the defendant in this securities case. Then predict the case outcome and award range."
Vertical Coverage
| Tier | Verticals | |------|-----------| | Standard | Personal Injury, Criminal Defense, Employment, Immigration, Civil Rights, ADR, Litigation Analytics, Game Theory, Appellate, Estate, Insurance, Legal Ethics, Class Action, Document Intelligence | | Premium | Family Law Deep, Tax, Real Estate, IP/Patent, Securities, Antitrust, Bankruptcy, Government Contracts, Construction, Healthcare, Environmental, Cybersecurity, Emerging Technology, Sports/Entertainment, International Trade, Judge Analytics, Firm Analytics |
Architecture
The standalone MCP server (splitifi-mcp) is a thin stdio client that forwards all tool calls to backend.mysplitifi.com/api/v1/intelligence via X-Splitifi-Key authentication. No data is stored locally.
For claude.ai integration (OAuth), use the backend-native MCP endpoint at https://backend.mysplitifi.com/api/v1/intelligence/mcp.
License
MIT. Predictions are informational — not legal advice. Always verify with licensed counsel in the relevant jurisdiction.
