Built for
Agents & Humans

The AlphaSwarm CLI and SDK provide a modernized, audited interface for full platform interaction. With the machine-readable Agent Contract, autonomous agents can discover and execute capabilities safely.

alphaswarm-cli roles assign
$ alphaswarm-cli roles assign [email protected] admin --scope-kind org --scope-id as-main --dry-run
Plan for roles assign:
+ verb: roles assign
+ method: POST
+ path: /tenancy/memberships
+ required_scopes: ["tenancy:admin"]
Propagations: eventual (on next login/refresh)

Modernized CLI Interface

Every command group (projects, deploy, services, roles, domains) follows a consistent 'emit' pattern, supporting JSON, Table, and YAML outputs for both human operators and automated scripts.

CLI Output Comparison (Table vs JSON)

Agent Contract

The machine-readable Agent Contract (JSON) defines every platform capability, required scopes, side-effects, and output schemas. Agents can parse the contract at runtime to understand the platform's surface.

Agent Contract Schema & Discovery Flow

Dry-Run & Plan

Safety first. Every mutating command supports --dry-run and --plan flags, emitting a structured plan object before execution. Agents can validate their intended actions against a policy engine before going live.

Dry-Run Plan Output & Policy Validation

Structured Error Handling

Standardized error codes (AswErrorCode) and machine-readable error envelopes ensure that agents can handle failures deterministically—from auth expiry to backend gaps.

Structured JSON Error Envelope

Developer Experience

Python SDK

A type-safe Python SDK for integrating platform capabilities directly into your agents and researchers.

Auto-Docs

Live documentation that is always in lock-step with the CLI tree, ensuring agents always have current information.

Dev Tooling

  • Modernized Typer-based CLI
  • Machine-Readable Agent Contract
  • Dry-Run & Plan Capabilities
  • Structured JSON/Table/YAML Output
  • Standardized Error Envelopes
  • mTLS Device Tunnels