Read Fireflies.ai meeting transcripts, summaries, soundbites, contacts, channels and analytics, upload audio, drive the notetaker in live meetings, and ask AskFred — over the Fireflies GraphQL API.
Fireflies.ai ships in the w6w first-party pack. It declares 25 actions, 3 health checks, and the host runs its code in a sandbox that never sees the credential.
io.w6w.firefliesFireflies.ai brings AI meeting intelligence into a workflow — reading transcripts, summaries, soundbites and analytics from recorded meetings, and driving the notetaker while a meeting is happening live. Actions cover searching and reading full transcripts, creating and searching short audio/video highlights (“Bites”), sharing or revoking access to a whole meeting, managing meeting privacy and channel assignment, and joining or steering an active meeting’s notetaker in real time.
Live-meeting actions let a workflow add action items as a call unfolds and adjust the notetaker’s state mid-meeting, while AskFred actions let a workflow ask questions of an AI assistant grounded in a team’s own meeting history. Administrative actions cover users, contacts and channels, and audio can be uploaded directly for transcription without a live meeting at all.
Every read and write goes through Fireflies’ GraphQL API, with each action modeling one specific operation rather than exposing raw query access — so a workflow author gets validated, described parameters instead of having to know the underlying schema.
Three routes to the same 25 actions. The Workflow tab is generated from Fireflies.ai's own manifest and carries its real ids, so it is copy-pasteable; the Code and CLI examples are the same call for any action on any app, so every app-specific value in them is a blank you fill in.
active-meeting-list List meetings the notetaker is currently in. These are the meeting ids the live-meeting actions take.
analytics-get Team-level meeting and conversation analytics over a date range — talk/listen ratio, filler words, sentiment, monologues.
app-output-list List the outputs Fireflies AI Apps produced, for one meeting or one app. The full list behind a transcript's `apps_preview`.
askfred-ask Ask a natural-language question about one meeting or a filtered set of meetings, and start a thread.
askfred-continue Ask a follow-up question in an existing AskFred thread, keeping its context.
audio-upload Send a publicly downloadable audio or video URL to Fireflies for transcription.
bite-search List soundbites owned by you, by your team, or clipped from one transcript.
channel-list List the team's channels. The ids are what `meeting-channel-update` and the Channel ID filter on `transcript-search` expect.
contact-list List everyone the API key's owner has met with, and when they last met. No arguments — Fireflies returns the whole list.
live-action-item-create Ask Fred to file an action item during a meeting that is still running, from a natural-language prompt.
live-meeting-join Add the Fireflies bot to a meeting that is already running.
live-meeting-state-set Pause or resume the notetaker's recording in a live meeting. The bot stays in the call either way.
meeting-channel-update Move 1–5 transcripts into a channel. All-or-nothing: if one fails validation, none are moved.
meeting-privacy-update Change who can see a meeting transcript. Requires ownership or team admin.
meeting-share Share a meeting transcript by email with people outside the team. They do not need a Fireflies account.
meeting-share-revoke Take away a previously shared external user's access to a meeting transcript.
meeting-title-update Rename a meeting transcript. Requires team admin privileges.
transcript-delete Permanently delete a meeting transcript. Returns the deleted record — the only copy you will have.
transcript-get Fetch one meeting transcript by id, with optional summary, sentences and media.
transcript-search List meeting transcripts, filtered by keyword, date range, people or channel.
A workflow step names the app and the action, and the editor fills in the
connection when you pick one. This is the Step shape from the
workflow spec, carrying Fireflies.ai's real ids.
{
"manifestVersion": "2",
"name": "fireflies-example",
"steps": [
{
"id": "askfred-ask",
"uses": {
"app": "io.w6w.fireflies",
"action": "askfred-ask",
"connection": "conn_YOUR_CONNECTION_ID"
},
"with": {
"query": "<query>"
}
}
]
}askfred-ask active-meeting-list analytics-get app-output-list askfred-continue +20 more actions available
Every app-specific value here is a blank you have to fill in. An
app action is reached through the connection that authenticates it, so the
address is a connection id, not the app id — and connections belong to your account,
so a public page cannot know yours. Create one for Fireflies.ai, then fill in
the three blanks: conn_YOUR_CONNECTION_ID, the action key, and the
parameters that action declares. The call itself is real — the shape is transcribed
from the studio's own snippet builder, which prints the same kind of blanks — but
nothing in it is specific to Fireflies.ai. The Workflow tab is where this app's
real ids are.
npm install @w6w/sdkyarn add @w6w/sdkpnpm add @w6w/sdkdeno add npm:@w6w/sdkimport { W6wClient, isActionRun } from "@w6w/sdk";
// Reads W6W_BASE_URL and W6W_TOKEN from the environment when omitted.
const client = new W6wClient();
const envelope = await client.run({
urn: "conn_YOUR_CONNECTION_ID",
action: "transcript-get",
payload: {
transcriptId: "<value>",
// includeSummary: "<value>",
// includeSentences: "<value>",
// includeMediaUrls: "<value>",
// includeAnalytics: "<value>",
},
});
if (isActionRun(envelope)) console.log(envelope.value); npm install -g @w6w/cli w6w run conn_YOUR_CONNECTION_ID --action transcript-get --payload '{"transcriptId":"<value>"}' Give an AI agent Fireflies.ai — without giving it Fireflies.ai's credentials. One MCP endpoint exposes every app, function and workflow the caller is entitled to, as tools it can discover and run. Access is granted per team while we onboard.
One tool call{
"name": "w6w_invoke",
"arguments": {
"ref": "app:io.w6w.fireflies#askfred-ask",
"input": {
"query": "<query>"
}
}
}
Every tool names its target with a single ref. The
app: form above doesn't name a connection at all — the
host resolves which of the caller's Fireflies.ai connections to sign
with, and refuses rather than guesses when the answer is ambiguous.
The token is attached host-side, at the moment of the call. It is never a tool argument, never in the model's context, and never in a transcript — so a prompt injection has nothing to exfiltrate.
Tools are derived per end user from what that person has actually connected and is entitled to — not one shared bot identity carrying the union of everyone's access.
Multi-step work runs on the workflow engine and returns a run handle the agent can poll — retries, branching and state survive the conversation that started them.
Fireflies.ai's declared health checks are on the surface too, so an agent can tell "the vendor is down" from "your credential expired" before it burns a retry on either.
The MCP surface is part of the hosted platform. Fireflies.ai itself is MIT, and the runtime that executes it is source-available (FSL).
Fireflies.ai declares its own checks, so its health is a property of the app rather than something the host guesses at.
Unauthenticated POST to https://api.fireflies.ai/graphql. An `auth_failed` reply passes — it proves the endpoint parsed the request and the auth layer ran. Credential validity is the derived `auth:*` check's job.