@speakai/mcp-server 1.28.0 → 1.28.2

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.
Files changed (3) hide show
  1. package/README.md +6 -6
  2. package/dist/index.js +27 -27
  3. package/package.json +3 -2
package/README.md CHANGED
@@ -402,10 +402,10 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
402
402
  | Tool | Description |
403
403
  |---|---|
404
404
  | `get_signed_upload_url` | Get a pre-signed S3 URL for direct file upload |
405
- | `upload_media` | Upload media from a URL — a direct/public file URL, a pre-signed S3 URL, or a social/video page link resolved automatically (YouTube, TikTok, Instagram, X/Twitter, Facebook, Reddit, SoundCloud, Twitch, Dailymotion, Streamable, Snapchat, Pinterest, Tumblr, Bilibili, VK, OK.ru, Rutube). Vimeo and Loom page links are not supported. |
405
+ | `upload_media` | Import an audio or video file from a direct public URL, a `get_signed_upload_url` URL, or a supported page link (YouTube, TikTok, Instagram, X/Twitter, Facebook, Reddit, SoundCloud, Twitch, Dailymotion, Streamable, Snapchat, Pinterest, Tumblr, Bilibili, VK, OK.ru, Rutube). Vimeo and Loom page links are not supported. Bills the media duration. |
406
406
  | `upload_local_file` | Upload a local file directly from disk (stdio and CLI only, never on the remote server) |
407
407
  | `upload_and_analyze` | Upload media from a URL (direct file, or any social/video page link `upload_media` accepts — resolved automatically) and return its `media_id` immediately. Poll `get_media_status` until `processed`, then call `get_media_insights` for AI summaries. Set `mediaType` when the user has said which they want; leave it off otherwise and the server picks the best available track. |
408
- | `upload_and_analyze_batch` | Upload up to 25 URLs in one call, 5 at a time. Each URL is reported as uploaded or failed with its reason, so one bad link does not sink the batch. Use this instead of calling `upload_and_analyze` in a loop. |
408
+ | `upload_and_analyze_batch` | Upload up to 25 URLs in one call, 5 at a time. Each URL is reported as uploaded or failed with its reason, so one bad link does not sink the batch. |
409
409
  | `list_media` | List and search media files with filters, pagination, and optional inline data (transcripts, speakers, keywords) via `include` param |
410
410
  | `get_media_insights` | Get AI insights — topics, sentiment, summaries, action items |
411
411
  | `get_transcript` | Get full transcript with speaker labels and timestamps |
@@ -502,7 +502,7 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
502
502
  | `bulk_assign_automation_folders` | Set folder scope for multiple automations |
503
503
  | `run_automations` | Manually run automations against media now |
504
504
  | `delete_automation` | Permanently delete an automation |
505
- | `list_automation_apps` | List catalog apps (native + integrations) |
505
+ | `list_automation_apps` | List catalog apps: Speak built-ins plus third-party apps marked connected or not connected (read-only) |
506
506
  | `list_automation_triggers` | List catalog trigger types (optionally by app) |
507
507
  | `list_automation_actions` | List catalog action/step types (optionally by app) |
508
508
 
@@ -652,8 +652,8 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
652
652
  | `get_voice_agent` | Get a single voice agent by agentId |
653
653
  | `create_voice_agent` | Create a voice agent (name, personality, instructions, and voice are required) |
654
654
  | `update_voice_agent` | Update a voice agent; send only the fields you want to change |
655
- | `delete_voice_agent` | Permanently delete a voice agent and its questions, test suite, and share link |
656
- | `create_voice_agent_from_prompt` | Create a new agent from a plain-English description instead of filling in fields |
655
+ | `delete_voice_agent` | Delete a voice agent; its share link, questions, and test suite stop working (no restore tool) |
656
+ | `create_voice_agent_from_prompt` | Create a new agent from a plain-English description; an LLM generates its name, instructions, and settings |
657
657
  | `generate_voice_agent_config` | Run the same prompt-to-config generation against an existing agent |
658
658
  | `get_voice_agent_setup_guide` | Discovery helper: the build order, which tool covers each piece, common mistakes |
659
659
  | `list_voice_avatars` | List the video avatars available to attach to a voice agent |
@@ -711,7 +711,7 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
711
711
  | `get_voice_test_suite` | Get a voice agent's test suite (its scenarios and run settings) |
712
712
  | `update_voice_test_suite` | Create or update a voice agent's test suite (upserts; replaces scenarios) |
713
713
  | `generate_voice_test_suite` | Auto-generate a default test suite from the agent's instructions and knowledge base |
714
- | `start_voice_test_run` | Start a test run against a voice agent's suite |
714
+ | `start_voice_test_run` | Queue a test run of a voice agent's scripted test suite (creates a queued run record; no calls are placed) |
715
715
  | `get_active_voice_test_run` | Get a voice agent's currently active run (queued, running, or paused) |
716
716
  | `pause_voice_test_run` | Pause a running or queued test run |
717
717
  | `resume_voice_test_run` | Resume a paused test run |
package/dist/index.js CHANGED
@@ -1442,22 +1442,22 @@ function register(server, client) {
1442
1442
  registerSpeakTool(
1443
1443
  server,
1444
1444
  "upload_media",
1445
- `Upload media from a URL \u2014 a direct/public file URL, a pre-signed S3 URL, or a shareable social/video page link, which Speak resolves to the underlying media automatically. Supported page links: ${SUPPORTED_URL_SOURCES}. ${UNSUPPORTED_URL_SOURCES} Processing is asynchronous \u2014 after uploading, use get_media_status to poll until state is 'processed' (typically 1-3 minutes for audio under 60 min), then use get_transcript and get_media_insights to retrieve results. For a single call that handles everything, use upload_and_analyze instead. For local files, use upload_local_file.`,
1445
+ `Import an audio or video file into Speak AI from a URL and start transcription. Accepts a direct public file URL, a URL returned by get_signed_upload_url, or a page link from a supported platform, which the server resolves to the underlying media. Supported page links: ${SUPPORTED_URL_SOURCES}. ${UNSUPPORTED_URL_SOURCES} Requires an active subscription. Creates a media item, bills its duration against the workspace's minutes or credits, and fires the workspace's media.created webhook if one is registered. The request fails if the file exceeds the plan's size limit or its duration cannot be read, and no mediaId is returned if the balance is insufficient. Returns mediaId and state right away while processing continues in the background. Use get_media_status until state is 'processed', then get_transcript and get_media_insights.`,
1446
1446
  {
1447
1447
  name: import_zod2.z.string().min(1).describe("Display name for the media file"),
1448
- url: import_zod2.z.string().describe("Direct/public media file URL, pre-signed S3 URL, or a shareable social/video page link \u2014 page links are resolved to the underlying media server-side. See this tool's description for the platforms accepted. Pass the URL the user gave you as-is; do not try to convert it to a file URL first."),
1448
+ url: import_zod2.z.string().describe("Direct public media file URL, a URL returned by get_signed_upload_url, or a page link from a platform listed in this tool's description. Page links are resolved server-side, so pass the URL the user gave you as-is."),
1449
1449
  mediaType: import_zod2.z.enum([MediaType.AUDIO, MediaType.VIDEO]).optional().describe('Type of media: "audio" or "video". Send it whenever the user has told you which they want \u2014 if they called it an audio file, or asked for audio only, pass "audio"; if they called it a video, pass "video". Otherwise omit it and the server decides: it inspects the actual file for a direct URL, and picks the best track the platform offers for a page link. Do not guess from the URL, because sending a value stops the server inspecting the file, and a video imported as "audio" can never be analysed as video afterwards.'),
1450
1450
  description: import_zod2.z.string().optional().describe("Description of the media file"),
1451
- sourceLanguage: import_zod2.z.string().optional().describe('BCP-47 language code for transcription, e.g. "en-US" or "he-IL"'),
1451
+ sourceLanguage: import_zod2.z.string().optional().describe('BCP-47 language code for transcription, e.g. "en-US" or "he-IL". Omit to use the default language on the user profile. An unsupported code falls back to automatic detection instead of failing.'),
1452
1452
  tags: import_zod2.z.string().optional().describe("Comma-separated tags for the media"),
1453
- folderId: import_zod2.z.string().optional().describe("ID of the folder to place the media in"),
1454
- callbackUrl: import_zod2.z.string().optional().describe("Webhook callback URL for this specific upload"),
1453
+ folderId: import_zod2.z.string().optional().describe("ID of the folder to place the media in. If the folder is not found, the media goes to the workspace's first folder."),
1454
+ callbackUrl: import_zod2.z.string().optional().describe("URL that replaces the workspace webhook's destination for this media's webhook events. It takes effect only when the workspace already has an active webhook for the event; on its own it does not create a webhook or send anything."),
1455
1455
  fields: import_zod2.z.array(
1456
1456
  import_zod2.z.object({
1457
1457
  id: import_zod2.z.string().min(1).describe("Custom field ID"),
1458
1458
  value: import_zod2.z.string().min(1).describe("Custom field value")
1459
1459
  })
1460
- ).optional().describe("Custom field values to attach to the media")
1460
+ ).optional().describe("Custom field values to attach to the media. Field IDs that do not belong to the workspace are ignored without an error.")
1461
1461
  },
1462
1462
  {
1463
1463
  title: "Upload Media from URL",
@@ -1607,7 +1607,7 @@ function register(server, client) {
1607
1607
  speakers: import_zod2.z.array(
1608
1608
  import_zod2.z.object({
1609
1609
  id: import_zod2.z.string().min(1).describe(
1610
- `Which speaker to rename. Accepts the speaker's CURRENT label exactly as it appears in the transcript (e.g. "Speaker 0", "Vatsal Shah"), or its numeric id from insight.speakers[].id (e.g. "0"). Not a fixed identifier \u2014 it changes when the speaker is renamed.`
1610
+ `Which speaker to rename. Accepts the speaker's CURRENT label exactly as it appears in the transcript (e.g. "Speaker 0", "Jane Doe"), or its numeric id from insight.speakers[].id (e.g. "0"). Not a fixed identifier \u2014 it changes when the speaker is renamed.`
1611
1611
  ),
1612
1612
  name: import_zod2.z.string().min(1).describe("New display name to assign to the speaker")
1613
1613
  })
@@ -1934,7 +1934,7 @@ function register(server, client) {
1934
1934
  speakers: import_zod2.z.array(
1935
1935
  import_zod2.z.object({
1936
1936
  id: import_zod2.z.string().min(1).describe(
1937
- `Which speaker to rename, matched against every file in mediaIds. Use the speaker's CURRENT label (e.g. "Vatsal Shah"). Only safe when that label already identifies the same person in every file listed \u2014 a default label like "Speaker 1", and any numeric id, is a per-file position and means a different person in each file. Not a fixed identifier: it changes when the speaker is renamed.`
1937
+ `Which speaker to rename, matched against every file in mediaIds. Use the speaker's CURRENT label (e.g. "Jane Doe"). Only safe when that label already identifies the same person in every file listed \u2014 a default label like "Speaker 1", and any numeric id, is a per-file position and means a different person in each file. Not a fixed identifier: it changes when the speaker is renamed.`
1938
1938
  ),
1939
1939
  name: import_zod2.z.string().min(1).describe("New display name to assign to the speaker")
1940
1940
  })
@@ -4862,7 +4862,7 @@ function register10(server, client) {
4862
4862
  registerSpeakTool(
4863
4863
  server,
4864
4864
  "create_automation",
4865
- "Create a new automation rule using the V2 graph model (trigger + ordered steps). Fetch valid step/trigger options with list_automation_triggers / list_automation_actions if unsure. For inbound-webhook automations the response includes inboundWebhook.inboundUrl (where to POST payloads) \u2014 recommended flow: create, send a test payload to the URL with ?test=1, call get_inbound_webhook to see mappable payload tokens, then update_automation to wire tokens/fieldsMap.",
4865
+ "Create an automation: a trigger plus ordered steps (V2 graph model). The automation is active by default and then runs on its own every time its trigger fires. Depending on its steps, each run can send email to any address, post to the workspace's Slack, send HTTP requests to any webhook URL, fetch a file from a URL into Speak, run actions in connected third-party apps through Composio, and use AI credits. An inbound-webhook trigger creates a public URL that accepts payloads (returned as inboundWebhook.inboundUrl). A Composio app-event trigger subscribes to events on the connected third-party account. Valid trigger and step types come from list_automation_triggers and list_automation_actions. To map inbound-webhook payload fields, send a test payload to the URL with ?test=1, read the tokens with get_inbound_webhook, then call update_automation.",
4866
4866
  writeSchema,
4867
4867
  {
4868
4868
  title: "Create Automation",
@@ -4901,7 +4901,7 @@ function register10(server, client) {
4901
4901
  registerSpeakTool(
4902
4902
  server,
4903
4903
  "update_automation",
4904
- "Update an existing automation rule. This replaces the whole automation (name, trigger, and steps), so fetch the current values with get_automation first and pass them all back with your changes.",
4904
+ "Update an existing automation rule. This replaces the whole automation (name, trigger, and steps), so fetch the current values with get_automation first and pass them all back with your changes. The saved steps run automatically on later triggers and can send email, post to Slack, call webhook URLs, and run actions in connected third-party apps. Changing a Composio app-event trigger updates the subscription on that third-party account.",
4905
4905
  {
4906
4906
  automationId: import_zod11.z.string().min(1).describe("Unique identifier of the automation"),
4907
4907
  ...writeSchema
@@ -5028,7 +5028,7 @@ function register10(server, client) {
5028
5028
  registerSpeakTool(
5029
5029
  server,
5030
5030
  "run_automations",
5031
- "Manually run one or more automations against one or more media items now (outside the normal trigger).",
5031
+ "Manually run one or more automations against one or more media items now (outside the normal trigger). Only active automations run; inactive or unknown automation ids and unknown media ids are skipped without an error. The runs happen in the background and this returns only an acknowledgement, so check results with get_automation_runs. Every step executes for real: it can send email, post to Slack, call webhook URLs, run actions in connected third-party apps, and use AI credits.",
5032
5032
  {
5033
5033
  mediaIds: import_zod11.z.array(import_zod11.z.string().min(1)).min(1).describe("Media ids to run the automations against"),
5034
5034
  automationIds: import_zod11.z.array(import_zod11.z.string().min(1)).min(1).describe("Automation ids to run")
@@ -5148,7 +5148,7 @@ function register10(server, client) {
5148
5148
  registerSpeakTool(
5149
5149
  server,
5150
5150
  "test_automation",
5151
- "Run an automation once against one media item to see which way it branches. THIS HAS REAL SIDE EFFECTS: only Speak's own run notifications are suppressed \u2014 outbound webhooks fire, Composio actions run against the connected third party, and AI steps are billed. Ask the user before calling it on an automation that posts anywhere outside Speak. It also needs a mediaId and sends no webhook payload, so an inbound-webhook automation cannot be meaningfully tested this way \u2014 its payload tokens will resolve to empty. Returns a runId; read the result with get_automation_run.",
5151
+ "Run one automation once against one media item to see which way it branches. This is a real run, not a dry run, and it works on inactive automations too. Only the run's own status notifications are suppressed and translation steps are skipped. Notify steps still send their email or Slack message, outbound webhooks still fire, Composio actions still run against the connected third-party account, and AI steps use credits. Confirm with the user before testing an automation that sends anything outside Speak. It needs a mediaId and sends no webhook payload, so on an inbound-webhook automation the payload tokens resolve to empty. Returns a runId; read the result with get_automation_run.",
5152
5152
  {
5153
5153
  automationId: import_zod11.z.string().min(1).describe("Unique identifier of the automation to test"),
5154
5154
  mediaId: import_zod11.z.string().min(1).describe("Media item to run the automation against")
@@ -5288,7 +5288,7 @@ function register10(server, client) {
5288
5288
  registerSpeakTool(
5289
5289
  server,
5290
5290
  "list_automation_apps",
5291
- "List the apps available in the automation catalog (e.g. Speak native + connected integrations). Use to discover what triggers/actions exist before building an automation.",
5291
+ "List the apps in Speak's automation catalog: Speak's built-in apps plus, when enabled, third-party apps available through Composio, each marked connected or not_connected for this user. Read-only: the list comes from Speak's own catalog and the user's saved connections, and no third-party service is called. Use the app slugs with list_automation_triggers and list_automation_actions.",
5292
5292
  {},
5293
5293
  {
5294
5294
  title: "List Automation Apps",
@@ -5342,7 +5342,7 @@ function register10(server, client) {
5342
5342
  registerSpeakTool(
5343
5343
  server,
5344
5344
  "list_automation_actions",
5345
- "List the action/step types available in the automation catalog. Optionally filter by app.",
5345
+ "List the step types (actions) available in Speak's automation catalog, optionally for one app slug, plus the fields each filter or condition step can test. Read-only: served from Speak's own catalog and the user's saved connections, with no third-party calls. Listing an action does not run it.",
5346
5346
  {
5347
5347
  app: import_zod11.z.string().min(1).max(100).optional().describe("Filter actions to a specific app slug")
5348
5348
  },
@@ -6345,9 +6345,9 @@ ${JSON.stringify(uploadRes.data, null, 2)}` }],
6345
6345
  registerSpeakTool(
6346
6346
  server,
6347
6347
  "upload_and_analyze_batch",
6348
- `Upload several URLs in one call \u2014 the batch form of upload_and_analyze, for when someone hands you a list of links. Takes up to ${MAX_BATCH_URLS} URLs and starts at most ${MAX_BATCH_CONCURRENCY} at a time so a long list does not hammer the API. Each URL may be a direct/public file URL or a shareable social/video page link. Supported page links: ${SUPPORTED_URL_SOURCES}. ${UNSUPPORTED_URL_SOURCES} One URL failing does not stop the rest: every URL is reported individually as uploaded or failed, with its reason. Returns as soon as the uploads are accepted, so poll get_media_status per mediaId, or list_media on the folder, to follow processing. Prefer this over calling upload_and_analyze in a loop.`,
6348
+ `Import up to ${MAX_BATCH_URLS} audio or video URLs in one call. Each URL is imported the same way as upload_and_analyze, and transcription starts for each one. At most ${MAX_BATCH_CONCURRENCY} uploads run at once. Each URL may be a direct public file URL or a page link from a supported platform, which the server resolves to the underlying media. Supported page links: ${SUPPORTED_URL_SOURCES}. ${UNSUPPORTED_URL_SOURCES} Each accepted upload creates a media item and bills its duration against the workspace's minutes or credits. A failed URL does not stop the others, and the result lists every URL as uploaded (with its mediaId) or failed (with the reason). Returns once the uploads are accepted. Use get_media_status per mediaId, or list_media on the folder, to follow processing.`,
6349
6349
  {
6350
- urls: import_zod15.z.array(import_zod15.z.string().min(1)).min(1).max(MAX_BATCH_URLS).describe("The URLs to import, up to 25. Pass each one exactly as the user gave it; page links are resolved server-side. Duplicates are uploaded once."),
6350
+ urls: import_zod15.z.array(import_zod15.z.string().min(1)).min(1).max(MAX_BATCH_URLS).describe("The URLs to import, up to 25. Pass each one exactly as the user gave it; page links are resolved server-side. Exact duplicate URLs are sent once."),
6351
6351
  mediaType: import_zod15.z.enum([MediaType.AUDIO, MediaType.VIDEO]).optional().describe('Applies to every URL in the batch. Send it only when the user has said which they want for all of them \u2014 "audio" if they asked for audio only, "video" if they called them videos. Otherwise omit it and the server decides per URL. Mixed batches: leave it off, or split into two calls.'),
6352
6352
  folderId: import_zod15.z.string().optional().describe("Folder ID for every upload in the batch"),
6353
6353
  sourceLanguage: import_zod15.z.string().optional().describe('BCP-47 language code applied to every upload, e.g. "en-US"'),
@@ -7212,7 +7212,7 @@ function register16(server, client) {
7212
7212
  registerSpeakTool(
7213
7213
  server,
7214
7214
  "create_dashboard",
7215
- `Create an analytics dashboard. Only \`title\` is required \u2014 source defaults to the whole workspace and dateRange to last30days. Add widgets by listing their types (the MCP assigns ids and lays them out automatically), scope with source ({type:"folders",folderIds} | {type:"team"} | {type:"workspace"}) and dateRange ({preset}), and optionally group widgets into sections. Design guidance: lead with a narrative widget as the first widget; group sections by the QUESTION they answer, not by widget type; don't pad \u2014 every widget earns its place (aim for 4-16 widgets on a full build); if something can't be expressed by the widget catalog, put it in a narrative widget's focus instead of faking it. Call list_dashboard_widgets first for the widget catalog, config vocabulary, design rules, and full examples. Viewer settings (the settings input): ` + SETTINGS_RULES,
7215
+ `Create an analytics dashboard. Only \`title\` is required \u2014 source defaults to the whole workspace and dateRange to last30days. Add widgets by listing their types (the MCP assigns ids and lays them out automatically), scope with source ({type:"folders",folderIds} | {type:"team"} | {type:"workspace"}) and dateRange ({preset}), and optionally group widgets into sections. Design guidance: lead with a narrative widget as the first widget; group sections by the QUESTION they answer, not by widget type; don't pad \u2014 every widget earns its place (aim for 4-16 widgets on a full build); if something can't be expressed by the widget catalog, put it in a narrative widget's focus instead of faking it. Call list_dashboard_widgets first for the widget catalog, config vocabulary, design rules, and full examples. Creating a dashboard does not share it publicly; only share_dashboard creates a public link. If settings.feedback.sheetWebhookUrl is set, each Feedback submission made on the shared dashboard is posted to that external Google Apps Script URL. Viewer settings (the settings input): ` + SETTINGS_RULES,
7216
7216
  {
7217
7217
  title: import_zod17.z.string().min(1).max(60).describe("Dashboard name, max 60 chars (the only required field)"),
7218
7218
  ...specFields,
@@ -7223,7 +7223,7 @@ function register16(server, client) {
7223
7223
  readOnlyHint: false,
7224
7224
  destructiveHint: false,
7225
7225
  idempotentHint: false,
7226
- openWorldHint: false
7226
+ openWorldHint: true
7227
7227
  },
7228
7228
  async ({ title, description, source, dateRange, sections, widgets, ...metadata }) => {
7229
7229
  try {
@@ -7468,7 +7468,7 @@ var init_dashboards = __esm({
7468
7468
  preset: import_zod17.z.enum(DATE_RANGE_PRESETS).describe("One of: last7days | last30days | last3months | yearToDate | allTime")
7469
7469
  }).describe("Date range \u2014 strict preset only, no free-form start/end dates");
7470
7470
  settingsFieldIds = import_zod17.z.array(import_zod17.z.string());
7471
- SETTINGS_RULES = "Do not pass settings unless the user explicitly asks to change this dashboard's viewer settings. Never pass settings for a Foxtons dashboard unless explicitly asked; Foxtons moves by a server script. Saving any settings section moves that dashboard onto the settings flow immediately: its media pages use these groups and this Feedback setup from then on. Each section (fields, feedback) replaces that whole section when sent. Call get_dashboard first and resend every key of the section you change; a key left out resets to its default. Get field ids from list_fields. Ids that are not the company's fields are dropped when saving. When feedback.isEnabled is true, pass a non-empty feedback.fieldIds (score fields) rather than leaving it empty. Only set feedback.sheetWebhookUrl when the user gives the Apps Script URL.";
7471
+ SETTINGS_RULES = "Do not pass settings unless the user explicitly asks to change this dashboard's viewer settings. Saving any settings section moves that dashboard onto the settings flow immediately: its media pages use these groups and this Feedback setup from then on. Each section (fields, feedback) replaces that whole section when sent. Call get_dashboard first and resend every key of the section you change; a key left out resets to its default. Get field ids from list_fields. Ids that are not the company's fields are dropped when saving. When feedback.isEnabled is true, pass a non-empty feedback.fieldIds (score fields) rather than leaving it empty. Only set feedback.sheetWebhookUrl when the user gives the Apps Script URL.";
7472
7472
  dashboardSettingsSchema = import_zod17.z.object({
7473
7473
  fields: import_zod17.z.object({
7474
7474
  includeIds: settingsFieldIds.describe(
@@ -7516,7 +7516,7 @@ var init_dashboards = __esm({
7516
7516
  "Per feedback field: a short row label and the allowed score range, used for both the reviewer's score and the approver's score."
7517
7517
  ),
7518
7518
  sheetWebhookUrl: import_zod17.z.string().optional().describe(
7519
- "Google Apps Script web app URL that receives one row per submission. Only script.google.com addresses are posted to. Never shown to viewers."
7519
+ "External Google Apps Script web app URL. Speak posts one row per Feedback submission (call date, media link, scores, submitter name, notes) to it. Only https://script.google.com/macros/s/<id>/exec addresses are called; other values are saved but never called. Never shown to viewers."
7520
7520
  )
7521
7521
  }).optional()
7522
7522
  }).describe(
@@ -7787,7 +7787,7 @@ function register17(server, client) {
7787
7787
  registerSpeakTool(
7788
7788
  server,
7789
7789
  "delete_voice_agent",
7790
- "Permanently delete a voice agent. Requires the OWNER or ADMIN role. Irreversible: also removes its questions, test suite, and share link; past conversations are kept for record-keeping but are no longer reachable from this agent.",
7790
+ "Delete a voice agent. Requires the OWNER or ADMIN role. The agent is marked deleted and no tool can restore it: it stops appearing in list_voice_agents, and its share link, embedded widget, questions, and test suite stop working. Its stored data is not erased, past conversations stay in list_voice_conversations, and phone numbers assigned to it are not released.",
7791
7791
  { agentId: import_zod18.z.string().min(1).describe("ID of the voice agent to delete (from list_voice_agents)") },
7792
7792
  { title: "Delete Voice Agent", readOnlyHint: false, destructiveHint: true, idempotentHint: true, openWorldHint: false },
7793
7793
  async ({ agentId }) => {
@@ -7802,11 +7802,11 @@ function register17(server, client) {
7802
7802
  registerSpeakTool(
7803
7803
  server,
7804
7804
  "create_voice_agent_from_prompt",
7805
- "Create a new voice agent by describing it in plain English instead of filling in name/personality/instructions/voice yourself. Requires the OWNER or ADMIN role. Response always includes the new agentId, plus either the generated agent config, or needsFollowUp: true with a follow-up question if the prompt was too thin to act on \u2014 call generate_voice_agent_config again on that agentId with more detail (or manualInstructions) when that happens.",
7805
+ "Create a new voice agent from a plain-English description, via an LLM call. Requires the OWNER or ADMIN role. Saves a draft agent, then generates and saves its name, personality, instructions, chat settings, and default voice, speech-to-text, and LLM settings. Does not assign a phone number. The response includes the new agentId plus either the generated agent, or needsFollowUp: true with a followUpQuestion when the prompt is too vague. In that case the draft agent still exists with placeholder settings; call generate_voice_agent_config on that agentId with more detail.",
7806
7806
  {
7807
7807
  prompt: import_zod18.z.string().min(1).describe('Plain-English description of the agent to build, e.g. "a friendly dental clinic receptionist that books appointments and answers insurance questions".'),
7808
- name: import_zod18.z.string().optional().describe("Agent name. Defaults to one derived from the prompt if omitted."),
7809
- manualInstructions: import_zod18.z.string().optional().describe("Skip generation and use this as the agent's instructions verbatim.")
7808
+ name: import_zod18.z.string().optional().describe("Initial name for the draft agent. Replaced by the generated name when generation succeeds; kept only if the response asks a follow-up question."),
7809
+ manualInstructions: import_zod18.z.string().optional().describe("Requirements the generated instructions must include. Generation still runs; when this is sent, no follow-up question is returned even if the prompt is vague.")
7810
7810
  },
7811
7811
  { title: "Create Voice Agent From Prompt", readOnlyHint: false, destructiveHint: false, idempotentHint: false, openWorldHint: false },
7812
7812
  async (body) => {
@@ -7825,7 +7825,7 @@ function register17(server, client) {
7825
7825
  {
7826
7826
  agentId: import_zod18.z.string().min(1).describe("ID of the existing voice agent to generate config for (from list_voice_agents)"),
7827
7827
  prompt: import_zod18.z.string().min(1).describe("Plain-English description of what the agent should do."),
7828
- manualInstructions: import_zod18.z.string().optional().describe("Skip generation and set the agent's instructions to this verbatim.")
7828
+ manualInstructions: import_zod18.z.string().optional().describe("Requirements the generated instructions must include. Generation still runs; when this is sent, no follow-up question is returned even if the prompt is vague.")
7829
7829
  },
7830
7830
  { title: "Generate Voice Agent Config", readOnlyHint: false, destructiveHint: true, idempotentHint: false, openWorldHint: false },
7831
7831
  async ({ agentId, ...body }) => {
@@ -8002,9 +8002,9 @@ function register18(server, client) {
8002
8002
  registerSpeakTool(
8003
8003
  server,
8004
8004
  "start_voice_test_run",
8005
- "Start a test run against a voice agent's suite. Requires the OWNER or ADMIN role. Rejects with 409 if the agent has no test suite." + NOT_WIRED_NOTE,
8005
+ "Queue a test run of a voice agent's scripted test suite. Creates a run record in queued status and returns it; it does not place phone calls, start a conversation, or use credits. Requires the OWNER or ADMIN role. Returns 404 if the agent or its test suite does not exist, and 409 if the suite has no enabled scenarios or the agent already has a queued, running, or paused run." + NOT_WIRED_NOTE,
8006
8006
  { agentId: import_zod19.z.string().min(1).describe("ID of the voice agent (from list_voice_agents)") },
8007
- { title: "Start Voice Test Run", readOnlyHint: false, destructiveHint: false, idempotentHint: false, openWorldHint: false },
8007
+ { title: "Queue Voice Agent Test Run", readOnlyHint: false, destructiveHint: false, idempotentHint: false, openWorldHint: false },
8008
8008
  async ({ agentId }) => {
8009
8009
  try {
8010
8010
  const result = await api.post(`/v1/voice/testing/${agentId}/run`);
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@speakai/mcp-server",
3
- "version": "1.28.0",
3
+ "version": "1.28.2",
4
4
  "mcpName": "io.github.speakai/mcp-server",
5
5
  "description": "Official Speak AI MCP Server — capture meetings, search thousands of recordings, run async voice and video surveys, create clips, and automate workflows from your AI assistant.",
6
6
  "homepage": "https://docs.speakai.co/mcp",
@@ -26,6 +26,7 @@
26
26
  "sync:check": "tsx scripts/sync-plugin.ts --check",
27
27
  "verify:plugin": "tsx scripts/verify-plugin.ts",
28
28
  "verify:plugin:live": "tsx scripts/verify-plugin.ts --live",
29
+ "build:openai-zip": "tsx scripts/build-openai-zip.ts",
29
30
  "test:watch": "vitest",
30
31
  "test:coverage": "vitest run --coverage"
31
32
  },
@@ -37,7 +38,7 @@
37
38
  "claude",
38
39
  "model-context-protocol"
39
40
  ],
40
- "author": "Vatsal Shah",
41
+ "author": "Speak AI",
41
42
  "license": "MIT",
42
43
  "engines": {
43
44
  "node": ">=22"