@meetstream/cli 0.6.0 → 0.6.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -2,6 +2,15 @@
2
2
 
3
3
  All notable changes to `@meetstream/cli`.
4
4
 
5
+ ## 0.6.1 (2026-11-21)
6
+
7
+ ### Fixed
8
+
9
+ - `meetstream transcript` printed `Unknown:` lines with no text: it read `speaker` / `transcript`, but segments name the speaker in `participant.name` and hold the text in `words[].text`. It now reads this shape (and the flat shape), and skips segments with no text.
10
+ - Zoom help text: `bot create` says Zoom bots join through your Zoom app with an OBF or ZAK token (`--zoom-obf-url` / `--zoom-zak-url`). `--zoom-use-own-app` / `--no-zoom-use-own-app` stay and now say which `use_own_app` value they send; the text no longer says Zoom needs no setup.
11
+ - README and AGENTS.md state: "Inbuilt keys work in pipeline mode only: Deepgram (speech-to-text) and OpenAI (LLM and voice). Realtime mode runs on OpenAI and needs your own OpenAI key. Any other provider: bring your own key." `key_source: "meetstream"` works in pipeline mode only.
12
+ - Removed an internal ticket reference from a test comment.
13
+
5
14
  ## 0.6.0 (2026-11-21)
6
15
 
7
16
  Matches MeetStream API v1.10.0.
package/README.md CHANGED
@@ -10,14 +10,19 @@ meetstream bot create "https://meet.google.com/abc-defg-hij" -t deepgram
10
10
 
11
11
  ---
12
12
 
13
- ## Two ways to integrate
13
+ ## Three ways to integrate
14
14
 
15
- 1. **Bot API: meeting bots and recording.** Send a bot to Zoom, Google Meet or Microsoft Teams, then get recordings, transcripts, participants, chat and webhooks after the call, or live data during the call. This is the default.
16
- 2. **Agents API: voice agents in the meeting.** A voice agent (API name: MIA) rides on the bot and answers by voice, in the chat, or with tool calls. Use it when you want an agent that talks.
15
+ | Flow | What it is | Price per meeting hour |
16
+ |---|---|---|
17
+ | **1. Recording only** (Bot API) | Send a bot to Zoom, Google Meet or Microsoft Teams, then get recordings, transcripts, participants, chat and webhooks after the call, or live data during the call. This is the default. | Bot price |
18
+ | **2. Agents in meetings, your own stack** (recommended first for an agent) | **2a. Bridge server:** the bot streams live audio to your server, and your server sends audio, chat and images back; you run your own speech-to-text, LLM and voice, hosted or local. **2b. Your own agent as the brain:** Claude Code, Codex, Hermes, OpenClaw and others answer through `model.provider: "local_agent"` and the Agent Kit. | 2a: bot price only. 2b: bot price + $0.25 agent infrastructure. No model tokens charged in either. |
19
+ | **3. Managed agent infrastructure** (Agents API) | MeetStream hosts the voice agent (API name: MIA). **3a.** MeetStream's inbuilt keys. Inbuilt keys work in pipeline mode only: Deepgram (speech-to-text) and OpenAI (LLM and voice). Realtime mode runs on OpenAI and needs your own OpenAI key. Any other provider: bring your own key. **3b.** Your own provider keys. | Bot price + $0.25 agent infrastructure + the models you pick |
20
+
21
+ Bot price: $0.35 per recording hour ($0.25 at volume). The MeetStream transcription add-on is $0.10 per hour, waived with your own provider key. Guide: [Integration paths](https://docs.meetstream.ai/guides/introduction/integration-paths).
17
22
 
18
- Guide: [Two ways to integrate](https://docs.meetstream.ai/guides/introduction/integration-paths).
23
+ Want an agent but not sure which flow? Run your own stack (flow 2, any models, local ones included), or let MeetStream host it (flow 3, no infrastructure work).
19
24
 
20
- **Path 1, fastest path.** Decide first: notes after the call (post-call), or something during the call (live)?
25
+ **Flow 1, recording only.** Decide first: notes after the call (post-call), or something during the call (live)?
21
26
 
22
27
  ```bash
23
28
  # Post-call: bot, webhooks, then the transcript
@@ -29,7 +34,19 @@ meetstream bot create "https://zoom.us/j/123456789" -t deepgram_streaming --live
29
34
  meetstream bot raise-hand <bot_id> # POST /api/v1/bots/{bot_id}/send_command {"command":"raise_hand"}
30
35
  ```
31
36
 
32
- **Path 2, fastest path.** Save an agent config, then send a bot with its id. Nothing else is needed on the bot.
37
+ **Flow 2a, bridge server.** The bot opens two WebSockets to your server: live audio in (`--live-audio-ws`, sent as `live_audio_required`) and the meeting control socket (`--socket-ws`, sent as `socket_connection_url`) for the audio, chat, images and interrupts you send back. Your server runs its own speech-to-text, LLM and voice, hosted or local (for example Parakeet for local speech-to-text). Bot price only.
38
+
39
+ ```bash
40
+ meetstream bot create "https://meet.google.com/abc-defg-hij" \
41
+ --live-audio-ws wss://your-bridge.example.com/bridge/audio \
42
+ --socket-ws wss://your-bridge.example.com/bridge
43
+ ```
44
+
45
+ Starter: [MeetStream Bridge](https://github.com/meetstream-ai/meetstream-agent) (FastAPI, OpenAI Realtime by default; swap in your own or local models). Guides: [Bridge server architecture](https://docs.meetstream.ai/guides/websockets/bridge-server-architecture), [Real-time audio streaming](https://docs.meetstream.ai/guides/websockets/real-time-audio-streaming), [Meeting control patterns](https://docs.meetstream.ai/guides/websockets/meeting-control-patterns).
46
+
47
+ **Flow 2b, your own agent as the brain.** Run the [MeetStream Agent Kit](https://github.com/meetstream-ai/agent-kit) bridge, save an agent with `"model": {"provider": "local_agent", "local_agent": {"url": "https://..."}}` (`meetstream mia create -f`), check it with `meetstream mia check <id>`, then send a bot with `--agent-config <id>`. Bot price + $0.25 agent infrastructure; no model tokens charged for your agent's reasoning. Flow 2b is opt-in and needs the local agent feature on for your account. Meeting speech reaches an agent that has access to your computer and terminal (prompt injection), so keep the lowest access level (`readonly`) and the activation code on, and keep your computer, the agent and the bridge running for the whole call. Guide: [MIA screen sharing and custom agents](https://docs.meetstream.ai/guides/features/mia-screen-sharing-and-custom-agents).
48
+
49
+ **Flow 3, managed agent infrastructure.** Save an agent config, then send a bot with its id. Nothing else is needed on the bot.
33
50
 
34
51
  ```bash
35
52
  cat > agent.json <<'JSON'
@@ -47,19 +64,13 @@ meetstream mia create -f agent.json # prints agent_config_
47
64
  meetstream bot create "https://meet.google.com/abc-defg-hij" --agent-config <agent_config_id>
48
65
  ```
49
66
 
50
- **Voice agent setups and pricing** (per meeting hour):
51
-
52
- | Setup | What you do | Price |
53
- |---|---|---|
54
- | a. MeetStream's inbuilt keys (default) | `"key_source": "meetstream"` on each pipeline section, no `api_key`. Inbuilt keys cover OpenAI (LLM, speech and voice models) and Deepgram (speech-to-text) today. | Bot price + $0.25 agent infrastructure + the per-hour price of each model you pick (the dashboard agent editor shows the rate before the bot runs) |
55
- | b. Your own keys | Your provider keys (OpenAI, Anthropic, Deepgram, ElevenLabs and others) under Integrations; `"key_source": "user"` or leave it out | Bot price + $0.25 agent infrastructure; your provider bills you for model usage |
56
- | c. Your own agent | `"model": {"provider": "local_agent", "local_agent": {"url": "https://..."}}` with the [MeetStream Agent Kit](https://github.com/meetstream-ai/agent-kit); check it with `meetstream mia check <id>` | Bot price + $0.25 agent infrastructure; no model tokens charged for your agent's reasoning |
57
-
58
- Bot price: $0.35 per recording hour ($0.25 at volume). The MeetStream transcription add-on is $0.10 per hour, waived with your own provider key. Details: [Voice agent pricing and keys](https://docs.meetstream.ai/guides/mia/pricing-and-keys).
67
+ - **3a. MeetStream's inbuilt keys:** `"key_source": "meetstream"` on each pipeline section, no `api_key`. Inbuilt keys work in pipeline mode only: Deepgram (speech-to-text) and OpenAI (LLM and voice). Realtime mode runs on OpenAI and needs your own OpenAI key. Any other provider: bring your own key.
68
+ - **3b. Your own keys:** your provider keys (OpenAI, Anthropic, Deepgram, ElevenLabs and others) under Integrations; `"key_source": "user"` or leave it out. Your provider bills you for model usage.
69
+ - **Price:** bot price + $0.25 agent infrastructure + the per-hour price of each model you pick (the dashboard agent editor shows the rate before the bot runs). Details: [Voice agent pricing and keys](https://docs.meetstream.ai/guides/mia/pricing-and-keys) and the [MIA API guide](https://docs.meetstream.ai/guides/mia/mia-api-guide).
59
70
 
60
- If inbuilt keys are not on for your account yet, `mia create` fails with `MeetStream-managed keys are not available for <section>. Use your own key.` Switch that section to your own key (setup b) and run it again; do not loop.
71
+ If inbuilt keys are not on for your account yet, `mia create` fails with `MeetStream-managed keys are not available for <section>. Use your own key.` Switch that section to your own key (3b) and run it again; do not loop.
61
72
 
62
- Setup c is opt-in and needs the local agent feature on for your account. Meeting speech reaches an agent that has access to your computer and terminal (prompt injection), so keep the lowest access level (`readonly`) and the activation code on, and keep your computer, the agent and the bridge running for the whole call. Guide: [MIA screen sharing and custom agents](https://docs.meetstream.ai/guides/features/mia-screen-sharing-and-custom-agents).
73
+ A voice agent bot takes only `--agent-config`. Do not add `--socket-ws` or `--live-audio-ws` to it (the usual cause of a silent agent); flows 2a and 3 are separate bots.
63
74
 
64
75
  ---
65
76
 
@@ -154,8 +165,7 @@ meetstream transcript <bot_id> --wait
154
165
  --separate-video capture per-participant video streams
155
166
  --zoom-zak-url <url> Zoom: HTTPS endpoint returning a ZAK token (signed-in join)
156
167
  --zoom-obf-url <url> Zoom: HTTPS endpoint returning an OBF token (on-behalf-of join)
157
- --zoom-use-own-app Zoom: always join with your own Zoom app
158
- --no-zoom-use-own-app Zoom: always join with no setup (not with ZAK/OBF)
168
+ --zoom-use-own-app Zoom: send use_own_app=true (needs your connected Zoom app)
159
169
  --storage-id <id> save the recording to this storage destination of
160
170
  your workspace (default: the workspace default)
161
171
  --google-login-domain <domain> Google Meet: join signed in with an account from this
@@ -182,7 +192,7 @@ meetstream transcript <bot_id> --wait
182
192
  --max-recording-seconds <sec> max in-call recording seconds (default 14400)
183
193
  ```
184
194
 
185
- **Zoom needs no setup.** Your API key and the meeting link are enough. With neither Zoom flag, the API chooses: no setup, or your own Zoom app when the account has saved Zoom app credentials or a connected Zoom account. `--zoom-zak-url` and `--zoom-obf-url` need your own Zoom app.
195
+ **Zoom joins through your Zoom app.** Bots join through your Zoom app (Meeting SDK) with an OBF token (`--zoom-obf-url`, on behalf of a user who is in the meeting) or a ZAK token (`--zoom-zak-url`, a signed-in bot), the way Zoom asks Meeting SDK apps to join since 2 March 2026. Setup: [Zoom Marketplace app setup](https://docs.meetstream.ai/guides/app-integrations/zoom-marketplace-app-setup) · [Zoom authenticated bots (ZAK and OBF)](https://docs.meetstream.ai/guides/app-integrations/zoom-authenticated-bots).
186
196
 
187
197
  **Signed-in bots.** `--google-login-domain` / `--teams-login-domain` send the `google_meet` / `teams` block (`login_required: true` plus the domain, and `sign_in_email` / `strict_email` when you pin an account). Pass one domain flag, not both. The CLI warns on stderr (and still sends) if the meeting link host doesn't match the platform. Signed-in Teams bots show the Microsoft account's own name and picture, so `--name` / `--image-url` are not applied.
188
198
 
@@ -311,7 +321,8 @@ npm test # node --test, fully mocked (no network)
311
321
 
312
322
  ## Latest docs
313
323
 
314
- - [Two ways to integrate](https://docs.meetstream.ai/guides/introduction/integration-paths) · [CLI](https://docs.meetstream.ai/build-with-ai/meetstream-cli)
324
+ - [Integration paths](https://docs.meetstream.ai/guides/introduction/integration-paths) · [CLI](https://docs.meetstream.ai/build-with-ai/meetstream-cli)
325
+ - [Bridge server architecture](https://docs.meetstream.ai/guides/websockets/bridge-server-architecture) · [MIA screen sharing and custom agents](https://docs.meetstream.ai/guides/features/mia-screen-sharing-and-custom-agents)
315
326
  - [Create Bot payload reference](https://docs.meetstream.ai/api-reference/create-bot-payload-reference) · [Webhooks and events](https://docs.meetstream.ai/guides/webhooks/webhooks-and-events) · [Errors](https://docs.meetstream.ai/errors)
316
327
  - [MIA API guide](https://docs.meetstream.ai/guides/mia/mia-api-guide) · [Voice agent pricing and keys](https://docs.meetstream.ai/guides/mia/pricing-and-keys)
317
328
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@meetstream/cli",
3
- "version": "0.6.0",
3
+ "version": "0.6.1",
4
4
  "description": "Official MeetStream CLI: create meeting bots, fetch transcripts, and stream webhook events from your terminal. Zoom, Google Meet, Microsoft Teams, including signed-in bots.",
5
5
  "type": "module",
6
6
  "bin": {
package/src/api.js CHANGED
@@ -152,7 +152,7 @@ export class MeetStreamClient {
152
152
  if (tid) {
153
153
  try {
154
154
  const { data } = await this.getTranscriptById(tid, { raw });
155
- // Live API may wrap segments in a `message` key: { message: [ {speaker, transcript, ...} ] }
155
+ // Live API may wrap segments in a `message` key: { message: [ {participant, words, ...} ] }
156
156
  const segments = Array.isArray(data) ? data : (Array.isArray(data?.message) ? data.message : data);
157
157
  if (segments && (Array.isArray(segments) ? segments.length : true)) {
158
158
  return { transcript_id: tid, transcript: segments };
@@ -387,7 +387,7 @@ export function buildCreateBotPayload(opts) {
387
387
  if (opts.zoomZakUrl) payload.zoom.zak_url = opts.zoomZakUrl;
388
388
  if (opts.zoomObfUrl) payload.zoom.obf_url = opts.zoomObfUrl;
389
389
  }
390
- // Zoom join choice. Omitted = automatic: no setup, or the account's own Zoom app when it has one.
390
+ // use_own_app (API field): sent only when --zoom-use-own-app / --no-zoom-use-own-app is passed.
391
391
  if (typeof opts.zoomUseOwnApp === 'boolean') {
392
392
  payload.zoom = { ...(payload.zoom || {}), use_own_app: opts.zoomUseOwnApp };
393
393
  }
@@ -16,7 +16,7 @@ export function registerBotCommands(program, getClient) {
16
16
 
17
17
  bot
18
18
  .command('create')
19
- .description('Send a bot to a meeting (or schedule with --join-at). Zoom needs no setup: the API key and the link are enough')
19
+ .description('Send a bot to a meeting (or schedule with --join-at). Zoom bots join through your Zoom app with an OBF or ZAK token (--zoom-obf-url / --zoom-zak-url)')
20
20
  .argument('<meeting-link>', 'Zoom / Google Meet / Teams meeting URL')
21
21
  .option('-n, --name <name>', 'bot display name', 'MeetStream Bot')
22
22
  .option('--video', 'record video as well as audio (default: audio only)')
@@ -33,8 +33,8 @@ export function registerBotCommands(program, getClient) {
33
33
  .option('--zoom-zak-url <url>', 'Zoom: HTTPS endpoint on your server returning a ZAK token (join as a signed-in user)')
34
34
  .option('--zoom-obf-url <url>', 'Zoom: HTTPS endpoint on your server returning an OBF token (join on behalf of a user in the meeting)')
35
35
  .option('--zoom-obf', 'removed: use --zoom-obf-url')
36
- .option('--zoom-use-own-app', 'Zoom: always join with your own Zoom app (default: automatic, no setup unless the account has its own Zoom app)')
37
- .option('--no-zoom-use-own-app', 'Zoom: always join with no setup (cannot be combined with --zoom-zak-url / --zoom-obf-url)')
36
+ .option('--zoom-use-own-app', 'Zoom: send use_own_app=true (needs your connected Zoom app; the API returns 400 without one)')
37
+ .option('--no-zoom-use-own-app', 'Zoom: send use_own_app=false (cannot be combined with --zoom-zak-url / --zoom-obf-url)')
38
38
  .option('--storage-id <id>', 'save the recording to this storage destination of your workspace (default: the workspace default)')
39
39
  .option('--google-login-domain <domain>', 'Google Meet: join signed in with an account from this registered Workspace domain (see: meetstream logins google)')
40
40
  .option('--teams-login-domain <domain>', 'Teams: join signed in with an account from this registered Microsoft 365 domain (see: meetstream logins teams)')
package/src/output.js CHANGED
@@ -26,11 +26,34 @@ export function kv(pairs) {
26
26
  }
27
27
  }
28
28
 
29
- /** Render transcript segments (speaker + transcript fields) as readable text. */
29
+ const str = (v) => (typeof v === 'string' ? v.trim() : '');
30
+
31
+ /** Speaker of a transcript segment: participant.name, else the flat speaker field. */
32
+ export function segmentSpeaker(s) {
33
+ return str(s?.participant?.name) || str(s?.speaker) || 'Unknown';
34
+ }
35
+
36
+ /** Text of a transcript segment: words[].text joined with spaces, else the flat transcript field. */
37
+ export function segmentText(s) {
38
+ if (Array.isArray(s?.words)) {
39
+ const parts = s.words.map((w) => str(w?.text ?? w?.punctuated_word ?? w?.word)).filter(Boolean);
40
+ if (parts.length) return parts.join(' ');
41
+ }
42
+ return str(s?.transcript) || str(s?.text);
43
+ }
44
+
45
+ /**
46
+ * Render transcript segments as readable "Speaker: text" lines. Segments name the speaker in
47
+ * participant.name and hold the text in words[].text; the flat speaker/transcript shape also works.
48
+ * Segments with no text are skipped.
49
+ */
30
50
  export function transcriptToText(segments) {
31
51
  if (!Array.isArray(segments)) return typeof segments === 'string' ? segments : JSON.stringify(segments, null, 2);
32
52
  return segments
33
- .map((s) => `${s.speaker ?? 'Unknown'}: ${s.transcript ?? ''}`.trim())
53
+ .map((s) => {
54
+ const text = segmentText(s);
55
+ return text ? `${segmentSpeaker(s)}: ${text}` : '';
56
+ })
34
57
  .filter(Boolean)
35
58
  .join('\n');
36
59
  }