@speakai/mcp-server 1.20.0 → 1.21.1
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/README.md +4 -3
- package/dist/index.js +156 -6
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -242,7 +242,7 @@ For questions about data handling, see [speakai.co/privacy](https://speakai.co/p
|
|
|
242
242
|
|
|
243
243
|
## What you can do once installed
|
|
244
244
|
|
|
245
|
-
Speak AI ships
|
|
245
|
+
Speak AI ships 113 tools your AI assistant can call. You don't memorize them — Claude/ChatGPT pick the right ones based on what you ask. Examples by category:
|
|
246
246
|
|
|
247
247
|
| Ask | Tools used (auto) |
|
|
248
248
|
|---|---|
|
|
@@ -394,7 +394,7 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
|
|
|
394
394
|
| `SPEAK_ACCESS_TOKEN` | No | Auto-managed | JWT access token (auto-fetched and refreshed) |
|
|
395
395
|
| `SPEAK_BASE_URL` | No | `https://api.speakai.co` | API base URL |
|
|
396
396
|
|
|
397
|
-
### MCP Tools (
|
|
397
|
+
### MCP Tools (113)
|
|
398
398
|
|
|
399
399
|
<details>
|
|
400
400
|
<summary>Media (17 tools)</summary>
|
|
@@ -422,11 +422,12 @@ SPEAK_API_KEY=your-key npx @speakai/mcp-server
|
|
|
422
422
|
</details>
|
|
423
423
|
|
|
424
424
|
<details>
|
|
425
|
-
<summary>AI Chat (
|
|
425
|
+
<summary>AI Chat (13 tools)</summary>
|
|
426
426
|
|
|
427
427
|
| Tool | Description |
|
|
428
428
|
|---|---|
|
|
429
429
|
| `ask_ai_chat` | Ask AI questions about media, folders, or your whole workspace |
|
|
430
|
+
| `get_analysis_quote` | Check if a file can be analysed as audio/video and what it costs |
|
|
430
431
|
| `retry_ai_chat` | Retry a failed or incomplete AI Chat response |
|
|
431
432
|
| `get_chat_history` | List recent AI Chat conversations |
|
|
432
433
|
| `get_chat_messages` | Get full message history for conversations |
|
package/dist/index.js
CHANGED
|
@@ -3149,7 +3149,10 @@ function register7(server, client) {
|
|
|
3149
3149
|
"Pass mediaIds for specific files, folderIds for entire folders, or omit both to search across all media.",
|
|
3150
3150
|
"Use assistantType to get specialized responses (e.g., 'researcher' for academic analysis, 'sales' for deal insights).",
|
|
3151
3151
|
"To continue a conversation, pass the promptId from a previous response.",
|
|
3152
|
-
"Returns a promptId \u2014 save it to continue the conversation with follow-up questions."
|
|
3152
|
+
"Returns a promptId \u2014 save it to continue the conversation with follow-up questions.",
|
|
3153
|
+
"Set analysisMediaId + analysisInput to have the model listen to the audio or watch the video instead of",
|
|
3154
|
+
"reading the transcript alone. That is a premium feature and costs credits per hour of media \u2014",
|
|
3155
|
+
"call get_analysis_quote first to check eligibility and price."
|
|
3153
3156
|
].join(" ");
|
|
3154
3157
|
const askAiChatInputSchema = {
|
|
3155
3158
|
prompt: import_zod8.z.string().min(1).describe("The question or prompt to ask about the media"),
|
|
@@ -3166,7 +3169,9 @@ function register7(server, client) {
|
|
|
3166
3169
|
isIndividualPrompt: import_zod8.z.boolean().optional().describe("When true, processes each media file separately instead of combining context. Useful for comparing responses across files."),
|
|
3167
3170
|
fieldId: import_zod8.z.string().optional().describe("Scope the prompt to a single custom field"),
|
|
3168
3171
|
fieldIds: import_zod8.z.array(import_zod8.z.string()).max(10).optional().describe("Scope the prompt to multiple custom fields (max 10)"),
|
|
3169
|
-
filters: import_zod8.z.record(import_zod8.z.unknown()).optional().describe("Advanced filter object to scope which media the prompt runs over")
|
|
3172
|
+
filters: import_zod8.z.record(import_zod8.z.unknown()).optional().describe("Advanced filter object to scope which media the prompt runs over"),
|
|
3173
|
+
analysisMediaId: import_zod8.z.string().optional().describe("Media to analyse as audio/video rather than transcript. Must also appear in mediaIds, and must be sent together with analysisInput. Premium feature."),
|
|
3174
|
+
analysisInput: import_zod8.z.enum(["audio", "video"]).optional().describe("'audio' lets the model hear tone, pacing and delivery; 'video' also lets it see what is on screen. Omit for transcript-only, which is the default and costs nothing extra. 'transcript' is not a valid value here \u2014 omitting the field IS transcript-only.")
|
|
3170
3175
|
};
|
|
3171
3176
|
const askAiChatAnnotations = {
|
|
3172
3177
|
title: "Ask AI Chat",
|
|
@@ -3175,11 +3180,52 @@ function register7(server, client) {
|
|
|
3175
3180
|
idempotentHint: false,
|
|
3176
3181
|
openWorldHint: false
|
|
3177
3182
|
};
|
|
3183
|
+
const ANALYSIS_TIMEOUT_MS = 6 * 60 * 1e3;
|
|
3184
|
+
const refuse = (message) => ({
|
|
3185
|
+
content: [{ type: "text", text: `Error: ${message}` }],
|
|
3186
|
+
isError: true
|
|
3187
|
+
});
|
|
3188
|
+
const withAnalysisOutcome = async (payload, messageId) => {
|
|
3189
|
+
if (!messageId) return payload;
|
|
3190
|
+
try {
|
|
3191
|
+
const res = await api.get("/v1/prompt/messages", { params: { messageId } });
|
|
3192
|
+
const messages = res.data?.data?.messages ?? res.data?.data ?? [];
|
|
3193
|
+
const list = Array.isArray(messages) ? messages : [];
|
|
3194
|
+
const match = list.find((m) => m?.messageId === messageId) ?? list[0];
|
|
3195
|
+
const analysis = match?.analysis;
|
|
3196
|
+
if (!analysis) return payload;
|
|
3197
|
+
const downgraded = analysis.requested && analysis.used && analysis.requested !== analysis.used;
|
|
3198
|
+
return {
|
|
3199
|
+
...payload,
|
|
3200
|
+
analysis,
|
|
3201
|
+
...downgraded ? {
|
|
3202
|
+
analysisWarning: `Requested ${analysis.requested} analysis but the answer came from the ${analysis.used}. ` + (analysis.skippedReason ?? "No reason was given.")
|
|
3203
|
+
} : {}
|
|
3204
|
+
};
|
|
3205
|
+
} catch {
|
|
3206
|
+
return payload;
|
|
3207
|
+
}
|
|
3208
|
+
};
|
|
3178
3209
|
const askAiChatHandler = async (params) => {
|
|
3210
|
+
const body = params ?? {};
|
|
3211
|
+
const analysisMediaId = typeof body.analysisMediaId === "string" ? body.analysisMediaId.trim() : "";
|
|
3212
|
+
const analysisInput = typeof body.analysisInput === "string" ? body.analysisInput.trim() : "";
|
|
3213
|
+
if (Boolean(analysisMediaId) !== Boolean(analysisInput)) {
|
|
3214
|
+
return refuse(
|
|
3215
|
+
"analysisMediaId and analysisInput must be provided together. Pass both to analyse audio/video, or neither for a transcript-only answer."
|
|
3216
|
+
);
|
|
3217
|
+
}
|
|
3218
|
+
const mediaIds = Array.isArray(body.mediaIds) ? body.mediaIds.map(String) : [];
|
|
3219
|
+
if (analysisMediaId && !mediaIds.includes(analysisMediaId)) {
|
|
3220
|
+
return refuse(
|
|
3221
|
+
`analysisMediaId "${analysisMediaId}" must also appear in mediaIds. mediaIds is currently ${mediaIds.length ? JSON.stringify(mediaIds) : "empty"}.`
|
|
3222
|
+
);
|
|
3223
|
+
}
|
|
3179
3224
|
try {
|
|
3180
|
-
const result = await api.post("/v1/prompt", params);
|
|
3225
|
+
const result = analysisInput ? await api.post("/v1/prompt", params, { timeout: ANALYSIS_TIMEOUT_MS }) : await api.post("/v1/prompt", params);
|
|
3226
|
+
const payload = analysisInput ? await withAnalysisOutcome(result.data, result.data?.data?.messageId) : result.data;
|
|
3181
3227
|
return {
|
|
3182
|
-
content: [{ type: "text", text: JSON.stringify(
|
|
3228
|
+
content: [{ type: "text", text: JSON.stringify(payload, null, 2) }]
|
|
3183
3229
|
};
|
|
3184
3230
|
} catch (err) {
|
|
3185
3231
|
return {
|
|
@@ -3196,6 +3242,36 @@ function register7(server, client) {
|
|
|
3196
3242
|
askAiChatAnnotations,
|
|
3197
3243
|
askAiChatHandler
|
|
3198
3244
|
);
|
|
3245
|
+
registerSpeakTool(
|
|
3246
|
+
server,
|
|
3247
|
+
"get_analysis_quote",
|
|
3248
|
+
"Check whether a media file can be analysed as audio or video, and what it will cost, before running ask_ai_chat with analysisInput. Returns { eligible, credits, seconds } and, when not eligible, a plain-English reason \u2014 an unavailable file is a normal result here, not an error. This is the only check that accounts for both the account's premium opt-in and the server-wide switch, so call it before committing to an expensive run.",
|
|
3249
|
+
{
|
|
3250
|
+
mediaId: import_zod8.z.string().min(1).describe("Media file to price"),
|
|
3251
|
+
analysisInput: import_zod8.z.enum(["audio", "video"]).describe("'audio' to hear the recording, 'video' to also see it. Video costs substantially more because frames dominate."),
|
|
3252
|
+
modelId: import_zod8.z.string().optional().describe("Optional model id to price against. Omit for the workspace default.")
|
|
3253
|
+
},
|
|
3254
|
+
{
|
|
3255
|
+
title: "Get Analysis Quote",
|
|
3256
|
+
readOnlyHint: true,
|
|
3257
|
+
destructiveHint: false,
|
|
3258
|
+
idempotentHint: true,
|
|
3259
|
+
openWorldHint: false
|
|
3260
|
+
},
|
|
3261
|
+
async (params) => {
|
|
3262
|
+
try {
|
|
3263
|
+
const result = await api.get("/v1/prompt/analysisQuote", { params });
|
|
3264
|
+
return {
|
|
3265
|
+
content: [{ type: "text", text: JSON.stringify(result.data, null, 2) }]
|
|
3266
|
+
};
|
|
3267
|
+
} catch (err) {
|
|
3268
|
+
return {
|
|
3269
|
+
content: [{ type: "text", text: `Error: ${formatAxiosError(err)}` }],
|
|
3270
|
+
isError: true
|
|
3271
|
+
};
|
|
3272
|
+
}
|
|
3273
|
+
}
|
|
3274
|
+
);
|
|
3199
3275
|
registerSpeakTool(
|
|
3200
3276
|
server,
|
|
3201
3277
|
"retry_ai_chat",
|
|
@@ -3978,11 +4054,50 @@ var init_inbound_webhook_utils = __esm({
|
|
|
3978
4054
|
}
|
|
3979
4055
|
});
|
|
3980
4056
|
|
|
4057
|
+
// src/capabilities.ts
|
|
4058
|
+
async function multimodalCapability(api) {
|
|
4059
|
+
if (cached && cached.expiresAt > Date.now()) return cached.value;
|
|
4060
|
+
let value = "unknown";
|
|
4061
|
+
try {
|
|
4062
|
+
const res = await api.get("/v1/user/profile");
|
|
4063
|
+
const flag = res.data?.data?.isMultimodalAnalysis;
|
|
4064
|
+
if (typeof flag === "boolean") value = flag ? "enabled" : "disabled";
|
|
4065
|
+
} catch {
|
|
4066
|
+
}
|
|
4067
|
+
cached = { value, expiresAt: Date.now() + TTL_MS };
|
|
4068
|
+
return value;
|
|
4069
|
+
}
|
|
4070
|
+
var TTL_MS, cached, MULTIMODAL_DISABLED_MESSAGE;
|
|
4071
|
+
var init_capabilities = __esm({
|
|
4072
|
+
"src/capabilities.ts"() {
|
|
4073
|
+
"use strict";
|
|
4074
|
+
TTL_MS = 10 * 60 * 1e3;
|
|
4075
|
+
cached = null;
|
|
4076
|
+
MULTIMODAL_DISABLED_MESSAGE = "Audio and video analysis is not enabled for this account. The step would be saved but would silently run on the transcript only, so it is refused here. Remove analysisInput from the step, or contact Speak AI to enable the feature.";
|
|
4077
|
+
}
|
|
4078
|
+
});
|
|
4079
|
+
|
|
3981
4080
|
// src/tools/automations.ts
|
|
3982
4081
|
var automations_exports = {};
|
|
3983
4082
|
__export(automations_exports, {
|
|
3984
4083
|
register: () => register10
|
|
3985
4084
|
});
|
|
4085
|
+
function stepsRequestMediaAnalysis(steps) {
|
|
4086
|
+
if (!Array.isArray(steps)) return false;
|
|
4087
|
+
return steps.some((step) => {
|
|
4088
|
+
const magicPrompt = step?.magicPrompt;
|
|
4089
|
+
const input = magicPrompt?.analysisInput;
|
|
4090
|
+
return typeof input === "string" && (input === "audio" || input === "video");
|
|
4091
|
+
});
|
|
4092
|
+
}
|
|
4093
|
+
async function refuseUngatedAnalysis(api, steps) {
|
|
4094
|
+
if (!stepsRequestMediaAnalysis(steps)) return null;
|
|
4095
|
+
if (await multimodalCapability(api) !== "disabled") return null;
|
|
4096
|
+
return {
|
|
4097
|
+
content: [{ type: "text", text: `Error: ${MULTIMODAL_DISABLED_MESSAGE}` }],
|
|
4098
|
+
isError: true
|
|
4099
|
+
};
|
|
4100
|
+
}
|
|
3986
4101
|
async function withInboundWebhookInfo(api, responseData, automationId) {
|
|
3987
4102
|
try {
|
|
3988
4103
|
if (!automationId) return responseData;
|
|
@@ -4130,6 +4245,8 @@ function register10(server, client) {
|
|
|
4130
4245
|
},
|
|
4131
4246
|
async (body) => {
|
|
4132
4247
|
try {
|
|
4248
|
+
const refusal = await refuseUngatedAnalysis(api, body.steps);
|
|
4249
|
+
if (refusal) return refusal;
|
|
4133
4250
|
const result = await api.post("/v1/automations/", body);
|
|
4134
4251
|
let data = result.data;
|
|
4135
4252
|
if (isInboundWebhookTrigger(body.trigger)) {
|
|
@@ -4164,6 +4281,8 @@ function register10(server, client) {
|
|
|
4164
4281
|
},
|
|
4165
4282
|
async ({ automationId, ...body }) => {
|
|
4166
4283
|
try {
|
|
4284
|
+
const refusal = await refuseUngatedAnalysis(api, body.steps);
|
|
4285
|
+
if (refusal) return refusal;
|
|
4167
4286
|
const result = await api.put(`/v1/automations/${automationId}`, body);
|
|
4168
4287
|
let data = result.data;
|
|
4169
4288
|
if (isInboundWebhookTrigger(body.trigger)) {
|
|
@@ -4414,8 +4533,19 @@ var init_automations = __esm({
|
|
|
4414
4533
|
init_helpers();
|
|
4415
4534
|
init_client();
|
|
4416
4535
|
init_inbound_webhook_utils();
|
|
4536
|
+
init_capabilities();
|
|
4417
4537
|
TOKEN_SYNTAX_NOTE = "Token syntax (usable in fields marked 'tokens allowed'): {{trigger.payload.<path>}} reads the inbound webhook payload (dot paths and [n] array indices; paths are relative to trigger.childKey when set \u2014 discover valid paths with get_inbound_webhook after sending a test payload); {{step.<index>.<path>}} or {{step.<stepId>.<path>}} reads a previous step's output (speak-upload -> mediaId, magic-prompt -> answer, outbound-webhook -> status/response).";
|
|
4418
|
-
STEPS_DESCRIPTION =
|
|
4538
|
+
STEPS_DESCRIPTION = `Ordered array of graph steps (1-20). Each step is an object: { stepId: string (unique within the array), stepType: one of "speak-upload" | "magic-prompt" | "translation" | "filter" | "condition" | "notify" | "outbound-webhook" | "composio-action", dependsOn?: string[] (stepIds this step runs after), branch?: "true"|"false" (which outcome of an upstream condition step this step belongs to) } plus ONE config key matching stepType:
|
|
4539
|
+
- speak-upload -> speakUpload: { sourceMode: "url"|"file", sourceUrl (required when sourceMode="url"; tokens allowed \u2014 if the token resolves to an object, the first http(s) URL inside it is used), folderId (required, unless folderRouting.mode="dynamic" where it becomes the optional fallback), name? (tokens allowed, mixable with static text), language? (language code or token), fieldsMap?: { <customFieldId>: "<value>" } (writes payload values into Speak custom fields on the uploaded media; values are usually {{trigger.payload.<path>}} tokens \u2014 get field ids from list_fields), folderRouting?: { mode: "static"|"dynamic", sourceKey (payload key holding the destination folder name, required when dynamic), onNoMatch: "create"|"default" (create a folder named after the value, or fall back to folderId) } }
|
|
4540
|
+
- magic-prompt -> magicPrompt: { prompt (required unless fieldIds given, max 20000), title?, assistantType? ("general"|"researcher"|"marketer"|"sales"|"recruiter"|"custom", default "general"), assistantTemplateId? (required if assistantType="custom"), fieldIds?: string[] (max 10 \u2014 extract answers into these custom fields), analysisInput? ("transcript" (default) | "audio" | "video") \u2014 what the model receives. "audio" lets it hear tone and delivery, "video" also lets it see what is on screen; on a video file "audio" extracts the audio track first. Premium: requires the account's audio/video analysis opt-in, and costs credits per hour of media }
|
|
4541
|
+
- translation -> translation: { targetLanguage: region-qualified locale code, e.g. "es-ES", "fr-FR" (bare codes like "es" are rejected) }
|
|
4542
|
+
- filter -> filter: { logic: "AND"|"OR" (default "AND"), rules: [{ field, op, value? }] (1-20) } \u2014 the run continues only when the rules match, otherwise it stops silently
|
|
4543
|
+
- condition -> condition: same { logic, rules } shape as filter, but instead of stopping it routes: downstream steps marked branch:"true"/"false" run according to the outcome
|
|
4544
|
+
- notify -> notify: { channel: "in_app"|"email"|"slack", target?, message (required, tokens allowed) }
|
|
4545
|
+
- outbound-webhook -> outboundWebhook: { url (required, tokens allowed), method? ("GET"|"POST"|"PUT"|"PATCH"|"DELETE", default "POST"), headers?: { <name>: <value> }, bodyTemplate?: string | object (tokens allowed) }
|
|
4546
|
+
- composio-action -> composio: { app, action, connectedAccountId?, argsTemplate? } (Composio is currently behind a server flag and may be unavailable)
|
|
4547
|
+
Filter/condition rule fields depend on what flows into the step: MEDIA -> name|duration|sourceLanguage|tags|transcript|speakers or a custom field id; INSIGHT -> answer; inbound-webhook DATA -> any payload path (e.g. "contact.status"). Ops by field type \u2014 text: eq|neq|contains|ncontains|startsWith|exists; number: eq|neq|gt|lt|exists; array: contains|ncontains|exists ("exists" takes no value; gt/lt values are numbers).
|
|
4548
|
+
` + TOKEN_SYNTAX_NOTE;
|
|
4419
4549
|
TRIGGER_DESCRIPTION = `Trigger object (the automation's root). Always include triggerSlug. Supported shapes:
|
|
4420
4550
|
- Media analyzed in folder(s): { type: "folders", triggerSlug: "media_analyzed", folderIds: string[] (min 1) }
|
|
4421
4551
|
- Inbound webhook (receive external payloads): { type: "folders", triggerSlug: "inbound_webhook", webhookId? (from provision_inbound_webhook; omit to auto-provision a new one on create), childKey? (dot-path narrowing which part of the payload feeds the automation, e.g. "data") }. The create/update response includes inboundWebhook.inboundUrl \u2014 the public URL to POST payloads to.
|
|
@@ -5147,6 +5277,15 @@ function register14(server, client) {
|
|
|
5147
5277
|
const magicPrompt = { prompt: spec.prompt ?? "", assistantType: "general" };
|
|
5148
5278
|
if (spec.title) magicPrompt.title = spec.title;
|
|
5149
5279
|
if (spec.model) magicPrompt.modelId = spec.model;
|
|
5280
|
+
if (spec.analyse || spec.analyze) {
|
|
5281
|
+
const analysis = String(spec.analyse ?? spec.analyze);
|
|
5282
|
+
if (!["transcript", "audio", "video"].includes(analysis)) {
|
|
5283
|
+
throw new Error(
|
|
5284
|
+
`Step ${i + 1} (ai_chat): analyse must be "transcript", "audio" or "video", got "${analysis}"`
|
|
5285
|
+
);
|
|
5286
|
+
}
|
|
5287
|
+
if (analysis !== "transcript") magicPrompt.analysisInput = analysis;
|
|
5288
|
+
}
|
|
5150
5289
|
if (Array.isArray(spec.saveToFields) && spec.saveToFields.length) {
|
|
5151
5290
|
if (spec.saveToFields.length > 10) {
|
|
5152
5291
|
throw new Error(`Step ${i + 1} (ai_chat): saveToFields supports at most 10 fields`);
|
|
@@ -5209,6 +5348,15 @@ function register14(server, client) {
|
|
|
5209
5348
|
for (const spec of orTriggers) entries.push(await buildTrigger(spec, false));
|
|
5210
5349
|
body.triggers = entries;
|
|
5211
5350
|
}
|
|
5351
|
+
if (wireSteps.some((s) => {
|
|
5352
|
+
const input = s?.magicPrompt?.analysisInput;
|
|
5353
|
+
return input === "audio" || input === "video";
|
|
5354
|
+
}) && await multimodalCapability(api) === "disabled") {
|
|
5355
|
+
return {
|
|
5356
|
+
content: [{ type: "text", text: `Error: ${MULTIMODAL_DISABLED_MESSAGE}` }],
|
|
5357
|
+
isError: true
|
|
5358
|
+
};
|
|
5359
|
+
}
|
|
5212
5360
|
const result = automationId ? await api.put(`/v1/automations/${automationId}`, body) : await api.post("/v1/automations/", body);
|
|
5213
5361
|
const resolvedId = unwrapData(result.data)?.automationId ?? automationId;
|
|
5214
5362
|
const response = {
|
|
@@ -5406,6 +5554,7 @@ var init_workflows = __esm({
|
|
|
5406
5554
|
fs = __toESM(require("fs"));
|
|
5407
5555
|
path2 = __toESM(require("path"));
|
|
5408
5556
|
init_media_utils();
|
|
5557
|
+
init_capabilities();
|
|
5409
5558
|
init_inbound_webhook_utils();
|
|
5410
5559
|
CANONICAL_FILTER_FIELDS = /* @__PURE__ */ new Set([
|
|
5411
5560
|
"name",
|
|
@@ -5422,7 +5571,7 @@ var init_workflows = __esm({
|
|
|
5422
5571
|
]);
|
|
5423
5572
|
ID_PATTERN = /^[0-9a-f]{12}$/;
|
|
5424
5573
|
TRIGGER_SPEC_DESCRIPTION = 'What starts the automation. Object with:\n- on (required): "media_analyzed" | "inbound_webhook" | "field_updated"\n- folders: array of folder names or ids (required for media_analyzed; missing folders are created)\n- childKey: dot-path narrowing the webhook payload root, e.g. "data" (inbound_webhook only)\n- webhookId: reuse a webhook from provision_inbound_webhook (inbound_webhook only; omit to auto-provision)\n- watchFields: array of { field: name-or-id, values?: string[] } (required for field_updated \u2014 fires when the field changes; values restricts to specific new values)\n- matchLogic: "AND"|"OR" for combining multiple watchFields value matches (default OR)';
|
|
5425
|
-
STEP_SPEC_DESCRIPTION = 'Ordered actions. Each step is an object with a `do` key plus its options. String values may be literals, "payload.<path>" shorthand (converted to {{trigger.payload.<path>}} only when it is the ENTIRE value), or raw {{...}} tokens \u2014 inside longer text, write the full {{trigger.payload.<path>}} form.\n- { do: "filter", rules: [{ field, op, value? }], logic?: "AND"|"OR" } \u2014 continue only if rules match. Fields: media flows use name|duration|sourceLanguage|tags|transcript|speakers or a custom field name; webhook payloads use payload paths like "contact.status". Ops: eq|neq|contains|ncontains|startsWith|gt|lt|exists\n- { do: "branch", rules, logic? } \u2014 like filter but routes instead of stopping; later steps with runWhen: "true"|"false" only run on that outcome. NOTE: branch routing requires the server\'s DAG runner (feature-flagged); when it is off, steps run in order and runWhen markers are ignored \u2014 prefer filter for guaranteed gating\n- { do: "upload", source (URL or payload.<path>, required), name?, language? (e.g. "en-US"), folder? (name or id; created if missing), folderFromPayload? (payload key holding the destination folder name \u2014 dynamic routing), onNoFolderMatch?: "create"|"default", mapFields?: { <field name or id>: <value or payload.<path>> } (writes payload values into custom fields on the uploaded media) }\n- { do: "ai_chat", prompt? (required unless saveToFields given), title?, saveToFields?: [field names or ids] (max 10 \u2014 values are extracted into these custom fields; prompt may be omitted for extraction-only steps), model? (a Speak-supported LLM id, e.g. "gemini-2.5-flash", "claude-sonnet-4-6"; omit for the workspace default) }\n- { do: "translate", language: region-qualified code like "es-ES", "fr-FR" }\n- { do: "notify", message (required, tokens allowed), channel?: "in_app"|"email"|"slack" (default in_app; email currently falls back to an in-app notification), target? (reserved \u2014 not yet used for delivery) }\n- { do: "call_webhook", url (required), method?, headers?, body? (string or object template, tokens allowed) }\nSteps may also set runWhen (after a branch step). Composio app actions (Google Drive, Slack apps, \u2026) are not supported by this builder yet \u2014 use create_automation directly for those.';
|
|
5574
|
+
STEP_SPEC_DESCRIPTION = 'Ordered actions. Each step is an object with a `do` key plus its options. String values may be literals, "payload.<path>" shorthand (converted to {{trigger.payload.<path>}} only when it is the ENTIRE value), or raw {{...}} tokens \u2014 inside longer text, write the full {{trigger.payload.<path>}} form.\n- { do: "filter", rules: [{ field, op, value? }], logic?: "AND"|"OR" } \u2014 continue only if rules match. Fields: media flows use name|duration|sourceLanguage|tags|transcript|speakers or a custom field name; webhook payloads use payload paths like "contact.status". Ops: eq|neq|contains|ncontains|startsWith|gt|lt|exists\n- { do: "branch", rules, logic? } \u2014 like filter but routes instead of stopping; later steps with runWhen: "true"|"false" only run on that outcome. NOTE: branch routing requires the server\'s DAG runner (feature-flagged); when it is off, steps run in order and runWhen markers are ignored \u2014 prefer filter for guaranteed gating\n- { do: "upload", source (URL or payload.<path>, required), name?, language? (e.g. "en-US"), folder? (name or id; created if missing), folderFromPayload? (payload key holding the destination folder name \u2014 dynamic routing), onNoFolderMatch?: "create"|"default", mapFields?: { <field name or id>: <value or payload.<path>> } (writes payload values into custom fields on the uploaded media) }\n- { do: "ai_chat", prompt? (required unless saveToFields given), title?, saveToFields?: [field names or ids] (max 10 \u2014 values are extracted into these custom fields; prompt may be omitted for extraction-only steps), model? (a Speak-supported LLM id, e.g. "gemini-2.5-flash", "claude-sonnet-4-6"; omit for the workspace default), analyse?: "transcript" (default) | "audio" | "video" \u2014 what the model receives. "audio" lets it hear tone and delivery, "video" also lets it see the screen; on a video file "audio" extracts the audio track first. Premium: requires the account\'s audio/video analysis opt-in and costs credits per hour of media }\n- { do: "translate", language: region-qualified code like "es-ES", "fr-FR" }\n- { do: "notify", message (required, tokens allowed), channel?: "in_app"|"email"|"slack" (default in_app; email currently falls back to an in-app notification), target? (reserved \u2014 not yet used for delivery) }\n- { do: "call_webhook", url (required), method?, headers?, body? (string or object template, tokens allowed) }\nSteps may also set runWhen (after a branch step). Composio app actions (Google Drive, Slack apps, \u2026) are not supported by this builder yet \u2014 use create_automation directly for those.';
|
|
5426
5575
|
buildAutomationSchema = {
|
|
5427
5576
|
name: import_zod15.z.string().min(1).max(150).describe("Display name for the automation"),
|
|
5428
5577
|
trigger: import_zod15.z.record(import_zod15.z.unknown()).describe(TRIGGER_SPEC_DESCRIPTION),
|
|
@@ -6725,6 +6874,7 @@ var init_tool_names = __esm({
|
|
|
6725
6874
|
"get_live_meeting_transcript",
|
|
6726
6875
|
// prompt
|
|
6727
6876
|
"ask_ai_chat",
|
|
6877
|
+
"get_analysis_quote",
|
|
6728
6878
|
"list_prompts",
|
|
6729
6879
|
"get_favorite_prompts",
|
|
6730
6880
|
"toggle_prompt_favorite",
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@speakai/mcp-server",
|
|
3
|
-
"version": "1.
|
|
3
|
+
"version": "1.21.1",
|
|
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",
|