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 { CatalogOp, CatalogService } from '@/plugins/catalog'
import type { CommandTree } from '@/plugins/commands/registry'
import { definePlugin } from '@/kernel/plugin'
import { catalog } from '@/plugins/catalog'
import { http } from '@/plugins/http'
import { session } from '@/plugins/session'
import { opsTree } from './tree'
export type ListOptions = Readonly<{ includeInternal: boolean }>
export type TreeOptions = ListOptions & Readonly<{ fresh: boolean }>
export type OpsService = {
readonly commands: (opts: TreeOptions) => Promise<CommandTree>
}
function shown(
ops: Readonly<Record<string, CatalogOp>>,
opts: ListOptions,
): Readonly<Record<string, CatalogOp>> {
return Object.fromEntries(
Object.entries(ops).filter(([, op]) => opts.includeInternal || !op.internal),
)
}
export const ops = definePlugin({
name: 'ops',
needs: [catalog, http, session],
build: async (ctx): Promise<OpsService> => {
const catalogService = await ctx.get(catalog)
const httpService = await ctx.get(http)
async function loaded(): Promise<CatalogService> {
if (!catalogService.loaded) await catalogService.replace(await httpService.fetchCatalog())
return catalogService
}
async function refresh(): Promise<Readonly<Record<string, CatalogOp>>> {
await catalogService.replace(await httpService.fetchCatalog())
return catalogService.ops()
}
const commands: OpsService['commands'] = async (opts) =>
opsTree(shown(opts.fresh ? await refresh() : (await loaded()).ops(), opts))
return { commands }
},
})