Wisebot CLI and HTTP API

@wisebotai/cli installs the wisebot command-line tool. It is not limited to workflows: most commands call the web app’s HTTP API (/api/v1/…) for organization metrics, agents, activity, webhooks, and API keys. Only wisebot workflow trigger targets workflow automation webhooks (a different host and credentials).

For workflow triggers from HTTP or scripts, see Triggering workflows (HTTPS & CLI).

Install

Package name: @wisebotai/cli.

npm install -g @wisebotai/cli

pnpm add -g @wisebotai/cli

Alternatively: npx @wisebotai/cli, a project devDependency, or pnpm exec wisebot. If the package is not on the registry you use, ask your administrator how the CLI is distributed.

Environment variables (two different contexts)

• WISEBOTAI_BASE_URL — Web app origin without a path (e.g. https://app.example.com or http://app.wisebotai.com).
• WISEBOT_ACCESS_TOKEN / WISE_KEY — Dashboard JWT or organization API key (wb_live_…) for Authorization: Bearer on /api/v1/….
• WISEBOT_WORKFLOW_TRIGGER_ORIGIN — HTTPS origin for /workflows/trigger (scheme + host, no path). The CLI also accepts WISEBOT_CONVEX_SITE or CONVEX_SITE_URL if WISEBOT_WORKFLOW_TRIGGER_ORIGIN is unset (same “site” host your administrator uses for Convex HTTP actions).
• WISEBOT_WORKFLOW_SECRET — Per-workflow webhook secret from the workflow editor.

Run wisebot, wisebot --help, or wisebot -h with no subcommand to print the embedded help (same command list as below).

Command reference (full)

Every subcommand below is implemented in @wisebotai/cli (wisebot). Invalid combinations print Usage: to stderr and exit with code 1. Successful HTTP responses print JSON (pretty-printed where noted) to stdout. Workflow trigger prints the response body (JSON if parseable).

wisebot · wisebot --help · wisebot -h

Prints the built-in help text (environment variables + command list). Does not call the network.

wisebot status

Example

export WISEBOTAI_BASE_URL="https://app.example.com"
export WISE_KEY="wb_live_…"
wisebot status

wisebot logs --tail

Requirements: the second token must be exactly --tail. wisebot logs without --tail is invalid and prints usage.

Example

wisebot logs --tail

wisebot agents list

Example

wisebot agents list

wisebot api-keys list

Example

wisebot api-keys list

wisebot webhooks list

Example

wisebot webhooks list

wisebot workflow trigger <workflowId> [--data '<json>']

Origin resolution (ORIGIN): first non-empty env among WISEBOT_WORKFLOW_TRIGGER_ORIGIN, WISEBOT_CONVEX_SITE, CONVEX_SITE_URL (each trimmed, trailing slash removed).

Examples

export WISEBOT_WORKFLOW_TRIGGER_ORIGIN="https://your-deployment.convex.site"
export WISEBOT_WORKFLOW_SECRET="wf_…"
wisebot workflow trigger "jd7abc123xyz"

wisebot workflow trigger "jd7abc123xyz" --data '{"message":"from CLI"}'

For curl, Node, and Python copies of the same POST, see Triggering workflows.

Quick reference (all commands)

Dashboard workflow run (not in CLI): POST /api/v1/flow/{workflowId} with JWT or wb_live_ key — see TypeScript SDK.

HTTP “SDK” for dashboard APIs

Use fetch, axios, requests, or any HTTP client against WISEBOTAI_BASE_URL and /api/v1/…, with the same Bearer token as the CLI.

If you use TypeScript or JavaScript, prefer the official TypeScript SDK (@wisebotai/sdk) for typed helpers: create agents, stream chat over SSE (/api/chat/agent), playground runs, workflows.run(), and knowledge by filename—without re-implementing those endpoints by hand. For Python, see Python integration (HTTP) (requests / httpx examples for the same paths).

Two workflow trigger paths:

See Triggering workflows for webhooks; TypeScript SDK for client.workflows.run().

Example (Node / TypeScript, raw fetch):

const base = process.env.WISEBOTAI_BASE_URL!.replace(/\/$/, "");
const token = process.env.WISEBOT_ACCESS_TOKEN ?? process.env.WISE_KEY;

const res = await fetch(`${base}/api/v1/status`, {
  headers: { Authorization: `Bearer ${token}` },
});
const text = await res.text();
if (!res.ok) throw new Error(text || res.statusText);
const data = JSON.parse(text) as Record<string, unknown>;

Paths mirror what the CLI calls (/api/v1/status, /api/v1/agents, /api/v1/activity, …). Exact payloads are subject to your deployment version.

For workflow POST examples (different base URL and secret), use Triggering workflows.

Playground commands (JWT only)

Playground subcommands call /api/v1/playground/*. They require a dashboard JWT — wb_live_ API keys return 403.

export WISEBOTAI_BASE_URL=http://app.wisebotai.com
export WISEBOT_ACCESS_TOKEN="<dashboard-jwt>"

wisebot playground session create --agent <agentId> --force-new
wisebot playground sessions list --agent <agentId>
wisebot playground session resume --conversation <conversationId>
wisebot playground send --conversation <id> --session <id> --message "Hello"
wisebot playground wait --conversation <id> --timeout 120
wisebot playground messages --thread <threadId>
wisebot playground traces --run <runId>
wisebot playground stop --conversation <id>
wisebot playground approvals pending --run <runId>
wisebot playground approve --id <approvalId>
wisebot playground deny --id <approvalId>

Choosing what to use

Related

• Python integration (HTTP) — requests / httpx for every CLI path + workflow POST
• TypeScript SDK (@wisebotai/sdk) — Typed agents, SSE chat, playground, workflows.run(), knowledge by filename
• Getting started — Dashboard layout and organization context
• Triggering workflows (HTTPS & CLI) — Webhook URL, dashboard API, and POST body for Input steps
• Workflows overview — Building flows, schedules, and run history