OpenClaw Mechanicus Patch

Extends OpenClaw with a full-stack automation platform — agent orchestration, neural intelligence (BrainJar spiking neural network), ClawScript DSL, process management, and a unified dashboard. Includes a complete IG Trading integration as a reference implementation.

Install

Cross-platform (Node.js — recommended):

git clone https://github.com/JoeSzeles/openclaw-mechanicus-patches.git
cd openclaw-mechanicus-patches
node install-node.cjs

The Node installer auto-detects your OpenClaw install location and works on Windows, Mac, and Linux. It patches the gateway to load ig-local-api.mjs at startup, enabling all API endpoints on the raw gateway without requiring the CEO proxy.

Linux / Mac (shell):

bash install.sh /path/to/openclaw

Windows (PowerShell):

.\install.ps1 C:\path\to\openclaw

The installer backs up any existing files before overwriting them.

Uninstall

Restores all original files from the backup created during install.

bash uninstall.sh /path/to/openclaw

.\uninstall.ps1 C:\path\to\openclaw

What's Included

Core Platform

• Mechanicus Dashboard — unified control surface for agents, processes, neural systems, and automation
• Agent Orchestration — spawn, manage, and chain agents with task assignment, parallel execution, and inter-agent communication
• Process Manager — bot lifecycle management with health monitoring, auto-restart, and log aggregation
• ClawScript DSL — general-purpose scripting language with 80+ commands spanning automation, AI, data, networking, and neural control
• Canvas System — visual workspace for dashboards, editors, documentation, and interactive tools

Neural Intelligence (BrainJar)

• Spiking Neural Network — biologically-inspired LIF neurons (5,000 default, configurable 350–20,000), Poisson-encoded inputs, synaptic weight learning
• Custom Brain Instances — create named brains with independent architectures, data directories, and persistent profiles (~/.openclaw/brains/)
• Volume Antenna — 180 neurons in 7 sub-groups for pattern detection (flash events, absorption, divergence, momentum shifts)
• Mushroom Body — dense interneuron cluster for memory consolidation and per-context pattern persistence across restarts
• Training Modes — backtest (historical data with feedback), live (real-time inputs), auto-test (multi-cycle calibration)
• Brain Management — boot, stimulate, observe, train, save/load, create/destroy instances via API or ClawScript

ClawScript

A general-purpose DSL for building automations, strategies, data pipelines, and agent workflows. Compiles to JavaScript with a visual flow editor for drag-and-drop composition.

Command categories:

• Core — variables, loops, conditions, functions, error handling, wait
• AI & ML — AI_QUERY, AI_GENERATE_SCRIPT, RUN_ML, ANALYZE_LOG
• Data & Web — WEB_FETCH, FILE_READ/WRITE, CLAW_WEB, CLAW_PDF, CLAW_X, DATA_TRANSFORM
• Agent & Task — AGENT_SPAWN/CALL/PASS/TERMINATE, TASK_ASSIGN/CHAIN/PARALLEL, SKILL_CALL
• Brain / Neural — BRAIN_BOOT, BRAIN_STIMULATE, BRAIN_OBSERVE, BRAIN_FEEDBACK, BRAIN_TRAIN, BRAIN_SAVE/LOAD, BRAIN_CREATE/USE/LIST/DESTROY, BRAIN_STATUS
• Communication — CHANNEL_SEND, EMAIL_SEND, SAY_TO_SESSION, WAIT_FOR_REPLY, ALERT
• Trading — BUY, SELL, SELLSHORT, EXIT, CLOSE, TRAILSTOP, INDICATOR, OPTIMIZE
• System — CRON_CREATE/CALL, STORE_VAR/LOAD_VAR, INCLUDE, CHAIN, PUBLISH_CANVAS

Full reference: github.com/JoeSzeles/clawscript

IG Trading Integration

A complete trading system built on the Mechanicus platform, demonstrating what's possible with agent orchestration and neural intelligence applied to financial markets.

• 23 strategies — scalper, trend-following, mean-reversion, breakout, Donchian, grid, pairs, and more
• Batch backtester — run multiple instruments/strategies/timeframes with optimization memory
• AI calibration agent — Groq-powered multi-cycle parameter tuning
• Signal monitor — real-time alerts from strategy signals
• Trade Claw engine — live trade execution via IG REST API
• Neural Trading — BrainJar applied to market data with Cortex auto-trade, Volume Antenna pressure sensing, per-instrument pattern cards

Neural Trading

The Neural Trading system integrates BrainJar's spiking neural network into the dashboard as four sub-tabs:

Tab	Purpose
Neural Dashboard	Live data feed, charts, connection status, manual triggers
Brain (Neural Cortex)	Per-context pattern cards, auto-execution, training controls, calibration
Brain Config	Network architecture, timeframe presets, sensory assignments, mushroom body, backtest training
Dev Console	Real-time brain activity logs

Key features:

• LIF Spiking Network — 5,000 neurons (configurable 350–20,000), Poisson-encoded data, synaptic weight learning
• Volume Antenna — 180 neurons in 7 sub-groups detect flash events, dead cat bounces, absorption, divergence, falling knives
• Cortex Auto-Execute — autonomous decision management with signal confirmation, hold zones, emergency exit on flash events
• Mushroom Body — dense interneuron cluster for memory consolidation, per-context pattern persistence across restarts
• Training — backtest (historical data with P&L feedback), live (real-time with antenna pressure), auto-test (multi-cycle calibration)

Documentation: open the Docs link in the Neural Trading tab bar, or view neural-trading.html in the canvas directory.

Environment Variables

Copy .env.example to .env in your OpenClaw directory and fill in your values:

cp .env.example .env

Variable	Required	Description
GROQ_API_KEY	No	Groq API key for AI calibration and AI_QUERY commands
OPENCLAW_LOGIN_USER	No	Username for dashboard login protection
OPENCLAW_LOGIN_PASSWORD	No	Password for dashboard login protection
DATABASE_URL	No	PostgreSQL connection string (see Database Setup below)
IG_API_KEY	For trading	IG Trading API key
IG_USERNAME	For trading	IG account username
IG_PASSWORD	For trading	IG account password
IG_ACCOUNT_ID	No	IG account ID (auto-detected if omitted)
IG_ACCOUNT_TYPE	No	demo (default) or live

Local API (ig-local-api.mjs)

The patch injects a standalone ESM module (ig-local-api.mjs) into the OpenClaw gateway at startup. This provides a full API layer that works on the raw gateway (port 18789) without requiring the CEO proxy.

Supported Endpoints (43 total)

Session & Config:

Endpoint	Method	Description
/api/ig/config	GET	Get config (credentials masked)
/api/ig/config	POST	Update config (accepts {profiles:{...}} or {profile,profileName})
/api/ig/config/test	POST	Test connection for a specific profile
/api/ig/session	GET	Get session status
/api/ig/session/refresh	POST	Force session token refresh

Trading (proxied to IG REST API):

Endpoint	Method	Description
/api/ig/positions	GET	List open positions with snapshots
/api/ig/positions/open	POST	Open a new position (BUY or SELL)
/api/ig/positions/close	POST	Close a position by dealId
/api/ig/positions/update	PUT	Update stop/limit on a position
/api/ig/account	GET	Account balance and details
/api/ig/markets	GET	Search markets by term
/api/ig/markets/:epic	GET	Get market details for an epic
/api/ig/marketnavigation	GET	Market navigation tree
/api/ig/pricehistory/:epic	GET	Historical price data
/api/ig/workingorders	GET	List working orders
/api/ig/workingorders/create	POST	Create a working order
/api/ig/workingorders/update	PUT	Update a working order
/api/ig/workingorders/delete	DELETE	Delete a working order
/api/ig/activity	GET	Account activity (requires ?from= param)
/api/ig/history	GET	Transaction history
/api/ig/watchlists	GET	IG account watchlists
/api/ig/refresh-snapshots	POST	Refresh market snapshots for positions

Streaming (local-mode stubs — requires CEO proxy for live data):

Endpoint	Method	Description
/api/ig/stream/prices	GET	Price stream status
/api/ig/stream/status	GET	Stream connection status
/api/ig/stream/candles	GET	Candle data (REST polling fallback)
/api/ig/stream/candle-stats	GET	Candle statistics

Strategies & Scalper (local JSON file CRUD):

Endpoint	Method	Description
/api/ig/strategies	GET	List all strategies
/api/ig/strategies/global	POST	Update global strategy toggle
/api/ig/strategy-templates	GET	List strategy templates
/api/ig/watchedlist	GET	Watched instruments list
/api/ig/proofread	GET/PUT	Proofread configuration
/api/ig/scalper	GET	Scalper configuration
/api/ig/scalper/status	GET	Scalper engine status
/api/ig/scalper/strategies	GET	Scalper strategies
/api/ig/scalper/strategy-schemas	GET	Strategy parameter schemas
/api/ig/scalper/backtests	GET	Backtest results
/api/ig/scalper/optimization-memory	GET	Optimization memory
/api/ig/scalper/batch-backtest	GET	Batch backtest results
/api/ig/logs/scalper-trades	GET	Scalper trade logs

ClawScript:

Endpoint	Method	Description
/api/clawscript/strategies	GET	ClawScript strategies
/api/clawscript/results	GET	ClawScript results
/api/clawscript/scripts	GET	ClawScript scripts
/api/clawscript/templates	GET	ClawScript templates
/api/clawscript/ai/config	GET	ClawScript AI configuration
/api/clawscript/logbook	GET/POST	ClawScript logbook entries

Bots & Processes:

Endpoint	Method	Description
/api/bots	GET	Registered bots
/api/processes	GET	Running processes

Config API Compatibility

The POST /api/ig/config endpoint accepts two formats for backward compatibility:

UI format (from model-config.js):

{ "profiles": { "demo": { "apiKey": "...", "username": "..." } } }

Direct format:

{ "profileName": "demo", "profile": { "apiKey": "...", "username": "..." } }

Masked values (•••••••• or ****) are automatically detected and skipped to prevent overwriting real credentials when the UI sends back masked fields.

Database Setup (PostgreSQL)

The system uses PostgreSQL for storing strategies, backtest results, optimization memory, trade history, and candle data.

No database? No problem. When DATABASE_URL is not set, the system automatically falls back to CSV files stored in ~/.openclaw/db/. Everything works — strategies, backtests, trades, optimization memory, candle caching — just using local CSV files instead of PostgreSQL. This means you can run the full system without any database setup at all.

PostgreSQL is recommended for production use (better performance with large datasets, concurrent access, proper indexing), but CSV mode is fully functional for development, testing, and personal use.

All tables are created automatically on first startup — no manual SQL required.

Option 1: Neon (Free Cloud PostgreSQL)

Neon offers a free tier with 512 MB storage — more than enough.

1. Go to https://neon.tech and sign up (GitHub/Google login works)
2. Click New Project — give it a name like openclaw-mechanicus
3. Select the region closest to you and click Create Project
4. On the dashboard, find the Connection string — it looks like:

   postgresql://username:password@ep-cool-name-123456.us-east-2.aws.neon.tech/neondb?sslmode=require
   

5. Copy the full connection string
6. Open your .env file and set:

   DATABASE_URL=postgresql://username:password@ep-cool-name-123456.us-east-2.aws.neon.tech/neondb?sslmode=require
   

7. Start (or restart) OpenClaw Mechanicus — tables are created automatically

Option 2: Supabase (Free Cloud PostgreSQL)

Supabase offers a free tier with 500 MB and a full Postgres database.

1. Go to https://supabase.com and sign up
2. Click New Project, pick an org, set a database password, and choose a region
3. After the project is created, go to Project Settings > Database
4. Under Connection string > URI, copy the connection string
5. Replace [YOUR-PASSWORD] in the string with the password you set
6. Add to your .env:

   DATABASE_URL=postgresql://postgres:[YOUR-PASSWORD]@db.xxxxx.supabase.co:5432/postgres
   

Option 3: Local PostgreSQL

If you have PostgreSQL installed locally:

DATABASE_URL=postgresql://postgres:yourpassword@localhost:5432/openclaw

Create the database first:

createdb openclaw

Verifying the Connection

After setting DATABASE_URL, start the system with start-mechanicus.ps1 (Windows) or start-mechanicus.sh (Linux/Mac). Look for these log messages:

[startup] Database: configured
[startup] price_candles table ready

If you see Database: not configured (file-only mode), check that your .env file has the correct DATABASE_URL value.

What the Database Stores

Table	Purpose
scalper_config	Global engine settings (budget, drawdown limits)
scalper_strategies	Strategy configurations (instruments, indicators, parameters)
scalper_trades	Execution history with P&L
scalper_backtests	Individual backtest results
optimization_memory	Best parameters found per instrument/strategy/timeframe
price_candles	Historical OHLCV data cached from APIs
agent_backups	Agent state snapshots
agent_memory	Agent long-term memory
agent_daily_memory	Agent daily journals
agent_subconscious	Agent pattern recognition data

Known Limitations

• Price history / chart data may return exceeded-account-historical-data-allowance on IG demo accounts (rate limit, resets after a few hours)
• Market navigation returns 500 from IG API on some demo accounts
• Activity endpoint requires ?from= date parameter (e.g., ?from=2026-03-01T00:00:00)
• Streaming (Lightstreamer live prices) works in local mode — auto-connects after first IG auth, subscribes to configured instrument epics, falls back to hybrid REST polling if L1 subscription fails. Live account streaming also supported if live profile is configured
• ClawScript compile/run/AI execution requires the CEO proxy — local mode returns helpful stubs
• Scalper engine start/stop requires the CEO proxy — local mode manages config/strategies/trades locally