Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.
Python agentagentic-aiagentic-frameworkagentic-workflowaiautomationclaudedeepseekgenaigptllmlow-codemcpnextjsno-codeopenaipythonskillsworkflow
import type { CommandConstructor, CommandEffect } from './command'
import type { CommandExample } from './examples'
import type { JsonSchema } from '@/plugins/catalog'
import { argToken } from '@/util/flag-name'
import { BINARY } from '@/version/info'
import { commandExamples } from './examples'
export type CommandRow = {
id: string
usage: string
summary: string
effect?: CommandEffect
input: JsonSchema
positional: readonly string[]
examples: readonly CommandExample[]
}
// The op side of a descriptor: present once a command wraps a catalog op instead of a
// hand-written zod schema.
export type OpFacets = {
op: string
method: string
path: string
kind: string
bind: Readonly<Record<string, string>>
deprecated: boolean
options?: JsonSchema
pins?: Record<string, string | null>
}
export type Descriptor = CommandRow & Partial<OpFacets>
const FLAGS = '[flags]'
export function commandRow(ctor: CommandConstructor, path: readonly string[]): CommandRow {
const id = path.join(' ')
const args = ctor.positional.map(argToken)
const input = ctor.schema()
const row: CommandRow = {
id,
usage: [BINARY, id, ...args, FLAGS].join(' '),
summary: ctor.summary,
input,
positional: ctor.positional,
examples: commandExamples({ id, input, positional: ctor.positional }, ctor.examples),
}
return ctor.effect === undefined ? row : { ...row, effect: ctor.effect }
}