# Amalgix — Financial Filing and Contract Intelligence > Cross-model evidence pipeline for AI agents. > Specialized models handle extraction and verification separately — each optimized for its stage. > Up to 3.8x cheaper than calling frontier models (Opus 4.8, GPT-5.5) directly for equivalent output. > Core paid workflows: analyze_public_filing, review_contract_risks, and analyze_document. > Pay-per-request with USDC via x402. ## Public Paid Workflows ### analyze_public_filing ($0.03–$79.00) — Financial Filing Intelligence - Analyze public SEC filings by content, filingUrl, ticker, or CIK - Supports 10-K, 10-Q, and 20-F filings - Returns metrics, risk changes, contradictions, management tone, missing information, evidence, confidence, and not_investment_advice - Best for: AAPL 10-K style analysis, filing-by-URL workflows, agent research pipelines. ### review_contract_risks ($0.03–$79.00) — Contract Risk Intelligence - Extracts parties, dates, commercial terms, obligations, risk flags, missing clauses, unusual terms, evidence, confidence, and not_legal_advice - Contract raw text is not persisted by default; stored records use hashes, metadata, job status, and optional results - Best for: sample MSA/NDA review, procurement agreements, SaaS contracts, and human-review triage. ### analyze_document ($0.03–$79.00) — General Evidence Analysis - Crucible™ Evidence Engine for source-grounded document analysis - Returns claim, evidence, sourceRef, confidence, and verificationStatus - Compatibility tool for general evidence extraction. Prefer the dedicated filing/contract tools when the workflow is known. ## Free Tools ### estimate_cost (FREE) — Cost Preview - Preview x402 USDC cost before paying - No charge, no commitment ### health_check (FREE) — Service Status - Verify service health and engine availability - Check before dispatching paid calls ## MCP Compatibility Tools - extract_web, summarize, and delegate_bulk_translate remain available through MCP for compatibility. - They are not public REST products by default. Legacy REST can be re-enabled by the operator with AMALGIX_ENABLE_LEGACY_REST_TOOLS=true. - OCR/scanned-PDF handling is an internal document ingestion capability, not a standalone public endpoint. ## Pricing - $79 is a hard safety cap, not a flat fee; ordinary inline documents quote far lower. - Prices scale dynamically with document size and request complexity, starting from $0.03 for small filings. - Use estimate_cost before paid calls for the exact quote. ## Quick Start ```json { "mcpServers": { "amalgix": { "url": "https://amalgix.io/mcp", "transport": "streamable-http" } } } ``` ## Features - Cross-Model Pipeline: Specialized extraction + verification models deliver stronger evidence at lower cost - Financial Filing Intelligence: SEC ticker/CIK lookup, filing URLs, public content analysis, cited evidence - Contract Risk Intelligence: clauses, obligations, risk flags, unusual terms, privacy-safe persistence - Crucible™ Evidence Engine: source-grounded findings with confidence and verification status - Cost Efficiency: Up to 3.8x cheaper than Opus 4.8 or GPT-5.5 for pipeline-equivalent output - Dynamic Pricing: Scales with document size from $0.03 floor to $79 ceiling - Schema-Enforced Output: Structured JSON with evidence citations and XSS sanitization - Wallet Dashboard: EVM/Base login, API keys, recent calls, x402 events, and job status - Self-Healing: Autonomous error recovery with retries and cross-model fallbacks - Dual-Chain Payment: USDC on Base (EVM) or Solana (SVM) - Limitations: not investment advice and not legal advice ## Discovery - /openapi.json — Full OpenAPI 3.1.0 spec - /.well-known/x402 — x402 resource discovery - /.well-known/x402.json — Enriched x402 discovery - /.well-known/mcp.json — MCP discovery - /.well-known/mcp/server-card.json — Smithery registry card - /.well-known/agent-card.json — A2A protocol agent card - /.well-known/agents.json — API discovery for LLM orchestrators - /mcp — MCP Streamable HTTP endpoint - /llms-full.txt — Complete reference - /.well-known/identity.json — Official identity declaration ## Contact - Website: https://amalgix.io - Email: support@amalgix.io