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 { CommandContext } from '@/plugins/base'
import { z } from 'zod'
import { assertNotEnvLogin, revokeAndClearSession } from '@/auth/logout'
import { pendingLoginStore } from '@/auth/pending-login'
import { Command } from '@/plugins/commands/command'
import { session } from '@/plugins/session'
const INPUT = z.object({})
const ENV_LOGOUT_MESSAGE = 'nothing to log out: the login comes from DIFY_TOKEN'
export default class Logout extends Command<typeof INPUT> {
static override summary = 'Log out of the current Dify session'
static override effect = 'write' as const
static override input = INPUT
async run(_input: z.infer<typeof INPUT>, ctx: CommandContext) {
const sessionService = await ctx.get(session)
// Refuse before touching anything: an env login must leave a pending interactive login alone.
assertNotEnvLogin(sessionService.fromEnv, ENV_LOGOUT_MESSAGE)
await (await pendingLoginStore(ctx)).clear()
const login = await sessionService.require()
await revokeAndClearSession(ctx, login)
return { logged_out: true, server: login.server }
}
}