Skip to content
PricingDocsBlog
Sign up →
⌘K
Getting started
  • Sign up and sign in
  • Workspaces and projects
  • Your first strategy
  • Your first backtest
Claude & ChatGPT
  • Connect Claude to Orbital
  • The eleven tools
    • get_authoring_guideread
    • create_strategywrite
    • build_programwrite
    • list_strategiesread
    • get_strategyread
    • update_strategywrite
    • estimate_backtestread
    • run_backtestrun
    • get_backtestread
    • get_backtest_tradesread
    • cancel_backtestdestructive
  • Rule cards and Programs
  • Troubleshooting
  • FAQ
  • ChatGPT
Chat
  • Chat
  • Ask about markets
  • Proposals and approvals
  • Documents, the web and your phone
Agents
  • Agents
  • Set up an agent
  • Permissions and approvals
  • Tasks
  • Automations
  • Channels and teams
  • Agents and strategies
Workflows
  • Workflows
  • Build a workflow
  • Node reference
  • Schedules
  • Webhooks and signals
  • Outputs, alerts and actions
  • Runs and versions
  • Troubleshooting workflows
Backtesting
  • Backtesting
  • Rule-card strategies
  • Program strategies in Rust
  • Run a backtest

Reference

  • Reading results
  • Cost profiles
  • Versions
  • Forward testing
  • Backtest, forward test and paper
  • When trades happen
  • Avoiding overfitting
  • Manage and delete strategies
  • Troubleshooting backtests
Trade
  • Trade on paper
  • Connect Alpaca
  • Deploy a strategy
  • Budgets and limits
  • Approvals and permissions
  • Manage a deployment

Reference

  • P&L, orders and activity
Watchlists
  • Watchlists
  • Notes and flags
  • Use watchlists in workflows and dashboards
Dashboards
  • Dashboards
  • Widgets
  • Templates
Alerts & notifications
  • Notifications
  • Alerts from workflows
Workspace & account
  • Profile and sign-in
  • Members and roles
  • Apps
  • Plugins
  • API keys
  • Credits and billing
  • AI models
  • Workspace settings
Glossary

Docs

Agents

Give an agent a task and the permissions to do it, then approve what it proposes.

  • AgentsWhat an Orbital agent is, how the Agents page is laid out, what an agent may do on its own, what it asks you for, and where its runs and results go.
  • Set up an agentCreate an agent with New agent, answer its short setup interview and apply the card, then adjust its model, instructions, permissions, tools and limits in Agent settings.
  • Permissions and approvalsWhat an agent may do on its own: the four permissions, the Deny, Ask, Auto and Always approve modes, who can change them, and how to answer the card an agent leaves while it waits.
  • TasksGive an agent, or a channel, a goal it keeps working at run after run: what done looks like, an optional budget, what to do when an idea fails, and the task bar that shows where it stands.
  • AutomationsStart an agent when nobody is here: on a schedule, with a task that repeats or starts on a market event, or when a monitor's signal fires.
  • Channels and teamsPut several agents in one channel, address one with @name or all at once, give the team a task the lead splits into sub-tasks, and bind the sources an agent can search.
  • Agents and strategiesWhat an agent does with strategies: reads them, writes and edits rule cards, branches, makes a version main, builds a program, runs backtests and controls a deployment, and what each act needs.

Orbital is software, not a broker or investment adviser. Results shown in Orbital are hypothetical and are not investment advice. Brokerage services are provided by the broker you connect. Terms · Privacy

From a sentence
to a filled order.

Orbital is in open beta. Paper trading on Alpaca.

Sign up →

Product

  • Overview
  • Chat
  • Agents
  • Workflows
  • Backtesting
  • Trade
  • Orbital MCPNew

Company

  • About
  • Blog
  • Careers
  • Press
  • Contact

Resources

  • Docs
  • Pricing
  • Discord ↗
  • RSS

© 2026 Orbital AI, Inc.

PrivacyTerms