# EarlyFra.me — Stateful Edge Signal & Action Orchestration Engine, 1-to-4 Router & MCP Server

> EarlyFra.me is a stateful edge webhook orchestration engine, plain-text alert rescuer, and Model Context Protocol (MCP) server. It turns 1 inbound webhook from TradingView, Stripe, Supabase, or GitHub into up to 4 simultaneous parallel outputs (Discord, Slack, Telegram, and Custom APIs with custom Authorization header injection) — and, unlike a stateless relay, it can wait for confirming signals, hold an execution for a human decision, mute itself during blackout windows, and retry durably.

Last Updated: 2026-10-19
Publisher: EarlyFra.me Systems Engineering (https://earlyfra.me/about)
Contact: curator@earlyfra.me

## Core Documentation & AEO Resources
- [EarlyFra.me Homepage & Interactive Simulacra](https://earlyfra.me/): Main application, Interactive Simulacra edge gate visualizer, unit economics calculator, and Multi-Route Studio.
- [Model Context Protocol (MCP) & Agentic Guide](https://earlyfra.me/docs/mcp): How to connect Claude Desktop, Cursor, Windsurf, and OpenAI Agents to EarlyFra.me via `https://relay.earlyfra.me/mcp`.
- [AI Agents & Bots Portal](https://earlyfra.me/agents): Machine-payable subscription quickstart — plan catalog, crypto checkout, claim endpoint, lifecycle callbacks, and agent MCP tool reference.
- [About EarlyFra.me & Edge Architecture](https://earlyfra.me/about): Engineering design, security model, and E-E-A-T credentials.
- [Contact & Security Disclosure](https://earlyfra.me/contact): Direct support, enterprise integration, and responsible disclosure via `curator@earlyfra.me`.
- [MCP Discovery Manifest](https://relay.earlyfra.me/.well-known/mcp.json): Machine-readable JSON-RPC 2.0 tool definitions for AI agents.
- [Product Updates RSS Feed](https://earlyfra.me/rss.xml): RSS 2.0 syndication feed.
- [Product Updates JSON Feed](https://earlyfra.me/feed.json): JSON Feed 1.1 syndication.

## Key Capabilities
1. **Plain-Text Alert Rescue:** Automatically parses raw non-JSON TradingView alerts (e.g., `SOLUSDT LONG @ 184.50 vol=4.2`) into structured JSON variables (`ticker`, `side`, `price`, `volume_spike`).
2. **Isolated Webhook Ingest Tokens (`whk_...`):** Public webhook ingest URLs use rotatable per-route `whk_...` tokens, keeping the user's private Master Passkey (`ef_live_...`) strictly isolated for Console and MCP login.
3. **Sliding-Window Deduplication & Noise Gate:** Drops duplicate bursts inside a 10s–3600s window in Edge Isolate RAM and evaluates multi-condition AND/OR rules against deep nested JSON paths (`data.object.amount_paid`).
4. **1-to-4 Parallel Fan-Out:** Dispatches formatted payloads simultaneously to Discord Rich Embeds, Slack Block Kit, Telegram Bot Markdown, and Custom HTTP/ERP endpoints with custom `Authorization: Bearer` header injection.
5. **Multi-Signal Confluence Traps (stateful_gate):** Buffer tagged signal legs (`timeframe`, `strategy`, …) for the same entity inside a sliding window; fire only when every `required_tags` leg has arrived, merging them into `payload._confluence_legs`. A per-entity cooldown then mutes repeats.
6. **Human-in-the-Loop Approval Gate (hitl_gate):** Hold the Orchestrator (Custom Bot / ERP) until a 1-click **Approve & Fire** / **Drop Signal** button inside the Discord, Slack, or Telegram message resolves a signed `GET /v1/approve/:approvalId` edge link. No bot token or callback registration required.
7. **Blackout Windows & In-Flight Enrichment (enrichment):** Auto-mute signals during UTC windows such as `13:55-14:15`, and merge a per-entity JSON lookup table into `payload._enriched` — readable from templates (`{{_enriched.risk}}`) and filter rules before evaluation.
8. **Durable Retry Queue & Time-Travel Replay:** 429/5xx/timeout failures are archived and retried by a cron-triggered sweeper with exponential backoff, and the most recent 25 inbound webhooks per passkey can be replayed against current rules as a backtest.
9. **Native MCP Server & Edge AI Enrichment:** Tier 3 (`Neural Frame`) exposes a JSON-RPC 2.0 MCP endpoint (`POST https://relay.earlyfra.me/mcp`) and injects real-time AI sentiment/summaries (`{{ai_summary}}`, `{{ai_sentiment}}`).
10. **Intelligent Schema Extraction & Template Preview:** `GET /api/v1/schema-catalog` (MCP `extract_schema_tokens`) walks the retained inbound webhooks and returns every `{{dot.path}}` the traffic actually fills — nested objects, array projections, inferred type, sample values and presence ratio — and `POST /api/v1/templates/preview` (MCP `preview_alert_template`) renders a draft title, body or JSON remapper against a real payload with zero quota. A `{{token}}` the payload does not carry is omitted from the delivered alert **together with the label before it**, so one ingest URL shared by several senders never leaks `RSI:` or a raw field name into the user's message.

## Transparent Flat-Rate Pricing
- **Free Sandbox Frame ($0):** 50 live webhooks (valid 7 days), up to 2 routes, unlimited zero-quota Dry-Run simulations.
- **Tier 1 — Pro Frame ($14.99/month or $149/year with 2 months free):** 250,000 monthly webhooks, up to 5 routes, 1-to-4 parallel fan-out, Custom Auth Header injection, JSON schema remapper.
- **Tier 2 — Apex Frame ($39.99/month or $399/year with 2 months free):** 1,000,000 monthly webhooks, up to 15 routes, 25 filter rules, 8.0s timeout, automatic 5xx/429 retry, and on-demand +500,000 Frame Boost Overage Packs ($19.99).
- **Tier 3 — Neural Frame ($79.99/month or $799/year with 2 months free):** 2,500,000 monthly webhooks, up to 50 routes, 50 filter rules, full Model Context Protocol (MCP) Server read/write access (`/mcp`), Programmatic REST API (`/api/v1/*`), Edge AI Signal Enrichment, and programmatic confluence-trap arming.