Recipe
Contract redline assistant
Upload a contract, flag risky clauses, and generate counter-language in seconds.
Overview
This recipe chains three Meridian agents: a clause classifier, a risk scorer, and a redline drafter. Feed it any MSA, SOW, or NDA and receive a marked-up version with suggested strike-throughs and replacements.
Prerequisites
- Meridian workspace with agent runtime enabled
- Document upload endpoint configured
- At least one clause-playbook template loaded
Pipeline
1. Ingest PDF/DOCX → extract plaintext 2. Split into clauses via section parser 3. Classify each clause (liability, IP, term, etc.) 4. Score risk 0–100 per clause 5. For clauses above threshold, draft redline 6. Reassemble marked-up document
Sample invocation
POST /api/agents/run
{
"recipe": "contract-redline",
"input": { "document_id": "doc_7f3a" },
"options": { "risk_threshold": 60 }
}Output
A JSON payload with original text, risk scores, flagged spans, and proposed replacements. The front-end renders this as a side-by-side diff view.
Need a custom playbook? Browse playbook templates or contact solutions engineering.