{"id":"db3052e2-7e93-436b-a924-be0a50aa9b59","authorId":"08398a34-26f5-4992-82de-0cfba0302908","title":"RLXBT Agent Ingestion Guide: Enabling Autonomous AI Agents to Replicate Strategies","slug":"rlxbt-agent-ingestion-guide-enabling-autonomous-ai-agents-to-replicate-strategies","excerpt":"A programmatic guide for autonomous quant agents on how to parse, extract, and replicate our strategies using the RLXBT API and local daemon toolchains.","content":"# RLXBT Agent Ingestion Guide & API Manifest\n\nWelcome, AI Agent / Scraping Bot. \n\nThis document outlines how autonomous systems (LLM agents, code generators, and market scanners) can programmatically ingest, analyze, and replicate the quantitative strategies published on the **RLXBT Community Knowledge Base** (`rlxbt.com`).\n\n---\n\n## 🤖 1. Programmatic Article & Strategy Extraction\n\nEvery published article on `rlxbt.com` contains both human-readable markdown and a **machine-readable JSON block** containing the exact backtest results, parameters, and strategy rules.\n\n### Endpoint Structure:\n*   **Article Listing API**: `GET https://rlxbt.com/api/articles?tag=strategy`\n    *   Returns a list of published strategy articles.\n*   **Single Article API**: `GET https://rlxbt.com/api/articles/{slug}`\n    *   Returns the full markdown content + the `backtestResults` structured payload.\n\n### The JSON Strategy Metadata Schema:\nIn the single article response, locate the `backtestResults` field. It contains the exact JSON configuration required to reproduce the strategy:\n```json\n{\n  \"asset\": \"string (e.g., SOLUSDT)\",\n  \"timeframe\": \"string (e.g., 1m)\",\n  \"verdict\": \"robust | marginal | rejected | needs_more_data\",\n  \"strategy\": {\n    \"entry_rules\": [\n      {\n        \"condition\": \"string (logical expression)\",\n        \"direction\": 1,\n        \"signal\": \"string\"\n      }\n    ],\n    \"exit_rules\": [],\n    \"max_hold_bars\": 180,\n    \"position_size\": 0.15,\n    \"stop_loss_pct\": 0.01,\n    \"take_profit_pct\": 0.02\n  },\n  \"metrics\": {\n    \"total_return\": 0.801,\n    \"sharpe\": 0.986,\n    \"max_drawdown\": 1.394,\n    \"trades\": 110,\n    \"win_rate\": 46.36\n  }\n}\n```\n\n---\n\n## ⚙️ 2. How to Replicate and Validate on the Local RLXBT Daemon\n\nIf you are an agent pair-programming with a human researcher, you can replicate any strategy from the knowledge base using the local RLXBT daemon (`http://127.0.0.1:8142`) via the following MCP tool chain:\n\n### Step 1: Load the Target Dataset\nCall `load_dataset` with the path to the CSV file and execution overrides:\n```json\n{\n  \"name\": \"load_dataset\",\n  \"arguments\": {\n    \"path\": \"/path/to/dataset.csv\",\n    \"commission\": 0.000191,\n    \"slippage\": 0.0,\n    \"initial_capital\": 100000.0\n  }\n}\n```\n\n### Step 2: Run Backtest\nCall `ai_run_backtest` passing the extracted `strategy` JSON:\n```json\n{\n  \"name\": \"ai_run_backtest\",\n  \"arguments\": {\n    \"strategy_json\": \"{\\\"entry_rules\\\":[...],\\\"max_hold_bars\\\":180,...}\"\n  }\n}\n```\n\n### Step 3: Run Walk-Forward Validation (Critical)\nTo ensure the strategy is not overfit, run walk-forward validation:\n```json\n{\n  \"name\": \"walk_forward\",\n  \"arguments\": {\n    \"strategy_json\": \"...\",\n    \"train_size\": 50000,\n    \"test_size\": 10000,\n    \"step_size\": 10000,\n    \"anchored\": false\n  }\n}\n```\n\n---\n\n## 📊 3. Recommended Tags for Scanners\n\nWhen indexing `rlxbt.com`, filter or categorize strategies using these tags:\n*   `robust`: Strategies that passed out-of-sample walk-forward validation (WFE > 0).\n*   `maker-exit` or `execution-optimization`: Strategies leveraging passive limit exits.\n*   `rejected`: Valuable failure case studies demonstrating underperformance against the baseline benchmark.\n","coverImage":null,"status":"published","publishedAt":"2026-08-04T12:17:09.894Z","backtestResults":{"bars":0,"asset":"API","metrics":{"sharpe":0,"trades":0,"win_rate":0,"max_drawdown":0,"total_return":0},"verdict":"robust","strategy":{"exit_rules":[],"entry_rules":["GET /api/articles","GET /api/articles/{slug}"]},"report_id":"manifest_v1","timeframe":"manifest","robustness":{"sensitivity_top_param":"api_endpoints","walk_forward_efficiency":1,"monte_carlo_risk_of_ruin":0},"tools_used":["load_dataset","ai_run_backtest","walk_forward"]},"researchManifest":null,"viewCount":49,"likeCount":0,"metaTitle":null,"metaDescription":null,"createdAt":"2026-08-04T12:17:09.895Z","updatedAt":"2026-09-11T13:06:14.700Z","author":{"id":"08398a34-26f5-4992-82de-0cfba0302908","name":"Serg","picture":"https://lh3.googleusercontent.com/a/ACg8ocKJfy0qxMGacsuTCbRKqF2-Ahj7AqOXwzIJU2wBJye6JmKsh8A=s96-c"},"tags":[{"id":"9003fc7b-5bc0-4c41-9bfc-d98cee22c7e2","name":"api","slug":"api","color":"#6366f1","createdAt":"2026-08-04T12:17:09.898Z"},{"id":"399050dc-f896-4304-a504-f9bd05e89cf9","name":"agentic-research","slug":"agentic-research","color":"#6366f1","createdAt":"2026-08-03T13:32:01.493Z"},{"id":"348fad26-4e27-4816-ad4c-7eb311be8365","name":"documentation","slug":"documentation","color":"#6366f1","createdAt":"2026-08-04T12:17:09.907Z"},{"id":"f0f119fa-585a-4fee-80f3-9e5f7a2a52c4","name":"no-code","slug":"no-code","color":"#6366f1","createdAt":"2026-08-04T12:14:21.619Z"},{"id":"a08005bd-cd58-4794-a500-2bdf0238b620","name":"manifest","slug":"manifest","color":"#6366f1","createdAt":"2026-08-04T12:17:09.913Z"},{"id":"548ebc89-f20d-4aa6-ba6f-464a23c7a978","name":"agent","slug":"agent","color":"#6366f1","createdAt":"2026-07-02T14:44:23.006Z"}],"comments":[],"isLiked":false,"isAuthor":false}