Integration

fluidX THE EYE integration

fluidX THE EYE (revXR) connects a service agent with a customer through the customer's phone camera: create a session and get an agent viewer link and an end-user link, invite the user by SMS or email, take photos remotely, annotate and download the captured photos and videos, read session usage, and close or delete sessions. Inbound calls let a kiosk or app request an available agent, poll until one accepts and hand the join link to the user. Sessions, contacts, photos and videos are personal data. Authentication is an API key generated in the fluidX dashboard (Admin, Generate API Key), sent in the x-api-key header; the role of the key's user decides which tools it may call.

Connect fluidX THE EYE How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in fluidX THE EYE.Built from the 21 fluidX THE EYE tools below, on the connection you add once.

What you can do with fluidX THE EYE

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Each badge shows the default agent permission of an endpoint (allowed: an agent calls it without asking; ask first: you are asked before an agent calls it, when the agent or chat asks permission for sensitive actions; never: no agent may call it), which every user can change endpoint by endpoint in their credential settings.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Session, 6 operations

close_all_sessionsClose all sessionsPOSTAgents: ask firstCloses every session matching status (only ACTIVE ones when left out) owned by the key's user; includeAll=true (TENANT_ADMIN keys) closes them for all users. Returns per-session outcomes. Answers 207 when some failed.
close_sessionClose sessionPOSTAgents: ask firstCloses one session; the viewer links stop working. Idempotent: a session already closing or closed returns its current state. Needs a USER role key or higher.
create_sessionCreate sessionPOSTAgents: ask firstCreates a TheEye session (live camera link between an agent and an end user) and returns sessionInfo.viewTheEyeUrl for the agent and sessionInfo.theEyeUserUrl for the user. Needs a PROJECT_ADMIN or higher key.
delete_sessionDelete sessionPOSTAgents: ask firstDeletes a session and everything captured in it, or only its photos and videos when onlymedia is true. Irreversible. Needs a TENANT_ADMIN or higher key.
get_session_infoGet session infoGETAgents: ask firstReturns the current state of one session: viewer links, session details, pixel streaming links and mediaInfo.photoRefs / videoRefs (ids to pass to download_media). The vendor allows only SUPER_ADMIN keys on it today.
list_user_sessionsList user sessionsGETAgents: ask firstLists the sessions of the key's user (all users with includeAll=true and a TENANT_ADMIN key), each with id, status, active flag, creation and expiry time and the session details. Filter by status.

Inbound call, 4 operations

cancel_inbound_callCancel inbound callPOSTAgents: ask firstCancels an inbound call that is not accepted yet (REQUESTED, ROUTING, RINGING). cancelled false means it is past that: use end_inbound_call. Idempotent.
create_inbound_callCreate inbound callPOSTAgents: ask firstAsks fluidX to route a TheEye call to an available agent. Returns callRequestId and callToken: pass both to get_inbound_call and poll every ~2 s until status is ACCEPTED or WAITING_FOR_USER, then send joinUrl to the user.
end_inbound_callEnd inbound callPOSTAgents: ask firstEnds an accepted inbound call (ACCEPTED, WAITING_FOR_USER, CONNECTED) and tears down the user's TheEye session. ended false means it cannot be ended: use cancel_inbound_call before acceptance. Idempotent.
get_inbound_callGet inbound callGETAgents: ask firstReturns the state of an inbound call; poll it every ~2 s. Terminal: ENDED, REJECTED, FAILED_NO_AGENT (no agent free), CANCELLED. joinUrl (the user's TheEye link) is set from ACCEPTED / WAITING_FOR_USER / CONNECTED.

Media, 2 operations

download_all_mediaDownload all mediaGETAgents: ask firstDownloads every photo and video of a session as one ZIP file (photos/ as PNG, videos/ in the tenant's default format) and returns it as a stored file. Needs a USER role key or higher.
download_mediaDownload mediaGETAgents: ask firstDownloads one captured photo (PNG) or video and returns it as a stored file. Needs a TENANT_ADMIN key. For a video, format mp4 falls back to webm when no MP4 version exists.

Media info, 2 operations

get_media_infoGet media infoGETAgents: ask firstReturns the title and description stored for one captured photo or video (mediaInfo, null when none was saved). Pass type image or video and the id from mediaInfo.photoRefs / videoRefs of get_session_info.
upsert_media_infoUpsert media infoPOSTAgents: ask firstCreates or replaces the title and description of one captured photo or video (type image or video, id from photoRefs / videoRefs). The caller must own the media or hold a TENANT_ADMIN key.

Photo, 2 operations

delete_photoDelete photoPOSTAgents: ask firstDeletes one captured photo by its numeric id (mediaInfo.photoRefs[].id). Irreversible. The vendor checks only the USER role, not who owns the photo.
take_photoTake photoPOSTAgents: ask firstMakes the agent's TheEye view take a photo from the user's live camera, as if the agent pressed Take Photo. Queued (202, status pending), run within about 2 s. The session must be ACTIVE with a running Eye session.

Email, 1 operation

send_emailSend emailPOSTAgents: ask firstSends a real email to a recipient, by default the TheEye invitation template (pass link to embed the session URL). subject and message override the content. The session must be ACTIVE. Needs a USER role key.

Media summary, 1 operation

get_media_summaryGet media summaryGETAgents: ask firstReturns, as one text, the filename, title and description of every annotated photo and video of a session (or a sentence saying there is no media). Use it to paste the annotations into a CRM.

Session consumption, 1 operation

get_total_consumption_duration_for_skuGet total consumption duration for skuGETAgents: ask firstReturns the total time a session consumed for one SKU: consumption.totalDurationInSeconds (raw, with milliseconds) and totalMinutes plus totalSeconds (rounded up). Only the SKU THEEYE is supported.

Sms, 1 operation

send_smsSend smsPOSTAgents: ask firstSends a real SMS to a phone number through seven.io with the account's sender ID, typically the session's user link. The session must be ACTIVE. Needs a USER role key or higher.

Video, 1 operation

delete_videoDelete videoPOSTAgents: ask firstDeletes one captured video by its numeric id (mediaInfo.videoRefs[].id). Irreversible. The vendor checks only the USER role, not who owns the video.

Use fluidX THE EYE from Claude Code, Cursor or Codex

Connect fluidX THE EYE in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its fluidX THE EYE operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect fluidX THE EYE

Create your workspace
Sign up free. Nothing to install.
Connect fluidX THE EYE once
Add your fluidX THE EYE connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.