iAgent Autopilot

Introducing Autopilot

The first AI Worker for autonomous Injective trading

Five agents handle surveillance, proposals, risk, execution, and audit — so your desk moves faster with hard limits and a receipt for every decision.

Launch workbenchInjective MCP
funding_flip · BTC· Injective perps

Read strategy policy and propose a funding reversion entry within max notional and leverage.

Injective MCPClaudeGroqWebSocketSQLite+ more

Skill

Funding rate crossed threshold — evaluate mean-reversion per strategy text and risk limits.

Analyzing market event…

Runs inside the stack you already trust

Injective MCPClaudeGroqWebSocketSQLiteFastAPINext.jsInjective MCPClaudeGroqWebSocketSQLiteFastAPINext.js

02 · Skills

Skilled, pre-trained & ready for Injective

Five specialized agents handle core trading work — surveillance through audit — with value visible in minutes on the live dashboard.

1 / 5

Skill

Market surveillance

Watcher ingests funding, drawdown, and breakout signals from the event bus and Injective feeds.

Watcher · Groq

03 · Outcomes

A new standard for autonomous trading ops

5 agents

Multi-agent orchestration

Supervisor-style pipeline: event → proposal → verdict → execution → audit.

100%

Decision receipts

Every chain is persisted and visible in the dashboard timeline — nothing is a black box.

1 write path

Governed execution

Only Executor touches chain writes. Kill switch and risk limits are enforced by default.

04 · How it works

AI that learns from real trading work

01

Text to trade plan

Describe strategy in natural language. Risk limits are parsed and saved — your team sets policy, Autopilot enforces it.

02

Events become skills

Each market event triggers the same five-agent pipeline. Resolutions are stored as auditable decision chains.

03

Autonomy with guardrails

Approved proposals execute via MCP. Rejections, kill switch, and dry-run keep human control built in.

05 · Platform

Powered by multi-agent orchestration

Built for hackathon demos and production guardrails — deploy on Render + Vercel without Docker.

Multi-agent orchestration

Work splits across Watcher, Analyst, Risk, Executor, and Auditor with coordinated handoffs.

Human-in-the-loop

Kill switch, risk veto, and strategy limits — safe autonomy without blind execution.

Governance & observability

API keys, audit stream, explorer links, and SQLite journal for every decision.

01 · Manifesto

Capital with conviction — on official Injective MCP

No parallel SDK. Every chain interaction flows through the Injective MCP Server. Architectural enforcement: only Executor submits writes. Risk, kill switch, and dry-run keep autonomy safe.