Documentation Index
Fetch the complete documentation index at: https://docs.archal.ai/llms.txt
Use this file to discover all available pages before exploring further.
What is a digital twin?
A digital twin is a stateful clone of a real service. It keeps track of objects, relationships, and constraints the same way the real service does.
Each twin:
- Implements the service-appropriate MCP and/or REST surface
- Runs an in-memory state engine with typed entity relationships
- Starts from a built-in seed or auto-generated state
- Exposes the service-appropriate hosted endpoints over HTTPS
- Runs on hosted infrastructure
Available twins
| Twin | What it covers | Details |
|---|
| GitHub | Repos, issues, PRs, branches, files, commits, actions, releases, code scanning | Hundreds of REST + MCP endpoints |
| Discord | Guilds, channels, messages, reactions, threads, webhooks, commands, interaction responses | REST-first bot workflows |
| Slack | Channels, messages, threads, reactions, users, files, admin, apps platform | Full Web API surface |
| Stripe | Customers, products, prices, payments, invoices, subscriptions, disputes, connect | Core billing and payments |
| Jira | Issues, projects, boards, sprints, fields, workflows, service desk, automation | Cloud REST API v3 |
| Linear | Issues, projects, teams, cycles, initiatives, roadmaps, labels, views | GraphQL API |
| Supabase | SQL queries, tables, migrations, extensions, edge functions, branches | PostgREST + Management API |
| Google Workspace | Gmail, Calendar, Drive, Sheets, Contacts | Google APIs |
| Ramp | Cards, funds, transactions, reimbursements, bills, approvals, travel | Ramp API |
| Telegram | Bot API — chats, messages, updates, inline queries | Bot API surface (MCP + REST) |
Compatibility matrix
Not every twin is exposed through every workflow. This is the canonical
reference — per-twin pages link back here.
| Twin | MCP | REST | Route mode (Vitest / --proxy) | State |
|---|
| GitHub | Yes | Yes | Yes | Stateful |
| Discord | Yes | Yes | Yes | Stateful |
| Slack | Yes | Yes | Yes | Stateful |
| Stripe | Yes | Yes | Yes | Stateful |
| Jira | Yes | Yes | Yes | Stateful |
| Linear | Yes | Yes | Yes | Stateful |
| Supabase | Yes | Yes | Yes | Stateful |
| Google Workspace | Yes | Yes | Yes | Stateful |
| Ramp | Yes | No | No | Stateful (MCP-only surface) |
| Telegram | Yes | Yes | No | Stateful |
api.github.com) and have the TLS proxy redirect traffic to the
twin with no code change. Ramp is MCP-only (no REST fidelity surface), and
Telegram has a route manifest gap today so it works via explicit base URL
injection only. All twins are stateful — there are no read-only twins.
Three ways to use twins
| Workflow | When to use it |
|---|
archal run | Run a scenario or task against twins, get a satisfaction score |
archal twin | Start persistent twins for debugging or manual integration work |
archal/vitest | Use supported hosted twins inside your existing Vitest test suite. Route mode currently supports Discord, GitHub, Google Workspace, Jira, Slack, Stripe, and Supabase |
Session lifecycle
archal run handles sessions automatically: start twins, run your agent, tear down.
archal twin start gives you a persistent session that stays alive for 30 minutes of inactivity. You can extend it with archal twin renew or tear it down with archal twin stop.
Seeds
Twins start from a seed, which is a predefined initial state. Each twin has built-in seeds for common configurations (e.g. github:small-project, github:enterprise-repo).
You can also describe the state you want in plain English in a scenario’s ## Setup section, and Archal generates the seed dynamically.
See the seeds guide for the full list of built-in seeds and how to use them.