Integration

Tess AI integration

Tess AI is an AI agent platform. List and inspect the agents a workspace can use, run one (with uploaded files and memory collections as context) or chat with it in the OpenAI format, read its executions, manage agent files, versions and webhooks, store memories in collections, upload and process files, read the workspace's usage history and its Enterprise audit events, and manage Digital Employees (scheduled autonomous agents): create, configure, run, pause, roll back, reparent and export them. Agent runs, file processing and employee runs spend the workspace's credits. Not offered: the streaming (SSE) routes and the large-file upload flow through a signed Google Cloud Storage URL. The API allows 1 request per second per organization (429 with retry_after above it). Authentication is a Tess API token sent as a Bearer token plus the numeric Workspace ID, sent as the x-workspace-id header on every call.

Connect Tess AI How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Tess AI.Built from the 60 Tess AI tools below, on the connection you add once.

What you can do with Tess AI

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.

Digital employee, 12 operations

create_employee_from_agentCreate employee from agentPOSTAgents: ask firstCreates a digital employee linked to an existing agent (agentId). With heartbeat_cron it wakes on that schedule and spends workspace credits on its own.
create_standalone_employeeCreate standalone employeePOSTAgents: ask firstCreates a digital employee with a new host agent. With heartbeat_cron it wakes on that schedule and spends workspace credits on its own; approval_required makes its runs wait for approval.
delete_digital_employeeDelete digital employeeDELETEAgents: ask firstSoft-deletes a digital employee: its data is kept and a workspace administrator can recover it.
detach_digital_employeeDetach digital employeePOSTAgents: ask firstRemoves a digital employee's parent, making it a top-level employee.
duplicate_digital_employeeDuplicate digital employeePOSTAgents: ask firstCreates a copy of a digital employee and returns the copy's id.
get_digital_employeeGet digital employeeGETAgents: ask firstGets a digital employee by id: id, name and status (for example idle, active, paused or pending_approval). 404 when it does not exist in the workspace.
list_digital_employeesList digital employeesGETAgents: ask firstLists the workspace's digital employees (id, name, status, linked agent_id, dates), 15 per page by default; filter by agent_id.
pause_digital_employeePause digital employeePOSTAgents: ask firstPauses a digital employee so no new scheduled run starts; reason is stored for audit. resume_digital_employee undoes it.
reparent_digital_employeeReparent digital employeePOSTAgents: ask firstMakes a digital employee the child of another employee in the same workspace (parent_employee_id); detach_digital_employee removes the parent.
resume_digital_employeeResume digital employeePOSTAgents: ask firstResumes a paused digital employee: its scheduled runs start again and spend workspace credits.
rollback_digital_employeeRollback digital employeePOSTAgents: ask firstRestores a digital employee's configuration to an earlier version (from list_employee_rollback_options), creating a new current version. Changes the live employee.
update_digital_employeeUpdate digital employeePUTAgents: ask firstChanges only the fields you send on a digital employee (name, status, modes, schedule, prompt, model, tools, knowledge files, delegation). Leave a field out to keep it. A schedule makes it run and spend credits.

File, 5 operations

delete_fileDelete fileDELETEAgents: ask firstDeletes a file by id (from list_files).
get_fileGet fileGETAgents: ask firstGets a file by id: filename, size in bytes, processing status (for example waiting or completed) and credits charged. Not cached: call again to follow processing.
list_filesList filesGETAgents: ask firstLists your Tess files (id, filename, size in bytes, processing status, credits charged), newest first by default, 15 per page, up to 100. Pass ids in execute_agent file_ids or link_agent_files.
process_fileProcess filePOSTAgents: ask firstProcesses an uploaded file by id and spends workspace credits; returns the file with its status and the credits charged.
upload_fileUpload filePOSTAgents: ask firstUploads one file (up to 32 MB: documents, spreadsheets, images, audio, video, code) to Tess and returns its id for execute_agent file_ids or link_agent_files. process true also processes it, spending credits.

Employee run, 4 operations

get_employee_runGet employee runGETAgents: ask firstGets one run of a digital employee: status (running, completed or failed), wake reason, start and finish times. Not cached: call again to follow a run started by invoke_digital_employee.
invoke_digital_employeeInvoke digital employeePOSTAgents: ask firstStarts a manual run of a digital employee now and spends workspace credits; returns the run (id, status running). Follow it with get_employee_run, then get_employee_run_output. 409 if a run is in progress.
list_employee_runsList employee runsGETAgents: ask firstLists a digital employee's runs (id, status running, completed or failed, wake reason, start and finish), 20 by default, up to 100; filter by status, page with offset. Not cached.
retry_employee_runRetry employee runPOSTAgents: ask firstRuns a finished run of a digital employee again and spends workspace credits; returns the new run. 409 if a run is in progress, 422 if the employee cannot run (paused or misconfigured).

Memory, 4 operations

create_memoryCreate memoryPOSTAgents: ask firstStores a memory (text up to 32,000 characters), optionally in a collection; agents executed with that collection in memory_collections retrieve it by semantic search.
delete_memoryDelete memoryDELETEAgents: ask firstDeletes a memory by id (from list_memories).
list_memoriesList memoriesGETAgents: ask firstLists the authenticated user's memories (id, text, collection_id), 10 per page by default, up to 50; filter by collection_id. Memories are free text the user stored and can hold personal information.
update_memoryUpdate memoryPATCHAgents: ask firstReplaces a memory's text (memory is required) and can move it to another collection; returns the memory with its collection.

Memory collection, 4 operations

create_memory_collectionCreate memory collectionPOSTAgents: ask firstCreates a memory collection and returns its id, under which create_memory stores memories and which execute_agent reads through memory_collections.
delete_memory_collectionDelete memory collectionDELETEAgents: ask firstDeletes a memory collection and all its memories. The default collection cannot be deleted (403).
list_memory_collectionsList memory collectionsGETAgents: ask firstLists the user's memory collections (id, name, display_name), 10 per page by default, up to 50; search filters them. Pass collection ids in memory_collections when executing an agent.
update_memory_collectionUpdate memory collectionPUTAgents: ask firstRenames a memory collection (name is required).

Webhook, 4 operations

create_agent_webhookCreate agent webhookPOSTAgents: ask firstAdds a webhook to an agent: Tess will call url (HTTPS only) with method POST or GET for that agent's notifications. No signing secret is documented, so deliveries cannot be verified.
delete_webhookDelete webhookDELETEAgents: ask firstDeletes a webhook by id (from list_webhooks or list_agent_webhooks); its URL stops receiving notifications.
list_agent_webhooksList agent webhooksGETAgents: ask firstLists the webhooks of one agent (id, url, method, status, delivery success and failure counts), 15 per page by default.
list_webhooksList webhooksGETAgents: ask firstLists webhooks across agents (id, template_id = agent id, url, method, status, delivery counts), 15 per page by default; list_agent_webhooks narrows to one agent.

Agent file, 3 operations

delete_agent_fileDelete agent fileDELETEAgents: ask firstDeletes a file associated with an agent (fileId from list_agent_files) and returns that file with removed true.
link_agent_filesLink agent filesPOSTAgents: ask firstLinks uploaded files (file_ids from upload_file or list_files) to an agent, associating them with it; returns the agent's files with counts by processing status (completed, failed, in progress).
list_agent_filesList agent filesGETAgents: ask firstLists the files linked to an agent (id, filename, size in bytes, processing status, credits charged) with the overall processing status. Not cached: call again to follow processing.

Agent version, 3 operations

get_agent_versionGet agent versionGETAgents: ask firstGets one saved version of an agent: its configuration snapshot (name, instructions, input questions, type, model settings, visibility, knowledge files) and who saved it (an email). Needs agent:version:read.
list_agent_versionsList agent versionsGETAgents: ask firstLists the saved versions of an agent with the fields each changed (change_summary, diff) and who saved it (published_by, an email), plus the current version. Needs the agent:version:read permission.
rollback_agent_versionRollback agent versionPOSTAgents: ask firstRestores an agent to a saved version by creating a new active version with that snapshot (history is kept). Changes the live agent. Rolling back to the current version is a 422.

Employee template pack, 3 operations

export_employee_template_packExport employee template packGETAgents: ask firstExports the digital employee configuration of an agent (agentId) as a template pack (version and employee settings); pass the whole answer as pack to preview_template_pack_import or import_template_pack.
import_template_packImport template packPOSTAgents: ask firstImports a template pack, creating or updating digital employees (a new host agent when target_agent_id is left out). collision_strategy overwrite replaces existing employees. Run preview_template_pack_import first.
preview_template_pack_importPreview template pack importPOSTAgents: ask firstDry run of a template pack import into target_agent_id: lists the employees that would be created, updated or skipped and any blocking issues. Changes nothing.

Agent, 2 operations

get_agentGet agentGETAgents: ask firstGets one Tess agent by id: title, type, visibility and questions, the input fields (name, type, required, options) execute_agent's input must carry. Call it before execute_agent or create_chat_completion.
list_agentsList agentsGETAgents: ask firstLists the Tess agents the workspace can use, 15 per page by default: id, title, type and questions (the inputs execute_agent takes). Filter with q (text search) or type. Public agents of other workspaces can appear.

Agent execution, 2 operations

execute_agentExecute agentPOSTAgents: ask firstRuns a Tess agent and spends workspace credits. Put the agent's fields (from get_agent) in input; with waitExecution true it waits up to 100 s, else poll get_agent_response with responses[0].id.
get_agent_responseGet agent responseGETAgents: ask firstGets an agent execution by id (responses[].id from execute_agent): status (starting while it runs, succeeded when done), output, credits charged and root_id to continue the thread. Not cached: call again to poll.

Employee hire, 2 operations

approve_employee_hireApprove employee hirePOSTAgents: ask firstApproves a digital employee waiting in pending_approval, making it active (it can then run and spend credits). Any other status is a 422.
reject_employee_hireReject employee hirePOSTAgents: ask firstRejects a digital employee's hire request in pending_approval status. Any other status is a 422.

Audit event, 1 operation

list_audit_eventsList audit eventsGETAgents: ask firstReads the workspace's audit event stream (Enterprise plans only) between from and to (30 days at most): actor (email, IP), entity, action, risk level and execution or credit metadata. Page with cursor.

Chat completion, 1 operation

create_chat_completionCreate chat completionPOSTAgents: ask firstSends a conversation to a Tess chat agent in the OpenAI chat completions format and returns the answer in choices[0].message.content. Spends workspace credits. Only messages, temperature, model and tools are read.

Digital employee capability, 1 operation

get_employee_capabilitiesGet employee capabilitiesGETAgents: ask firstTells what the token's user may do with digital employees in this workspace: create or update (can_write), start runs (can_invoke), pause, resume or roll back (can_manage_state). Check before writes.

Employee activity, 1 operation

get_employee_activityGet employee activityGETAgents: ask firstLists a digital employee's activity log (event name, who caused it by id and name, when), 50 events by default, up to 100.

Employee avatar, 1 operation

upload_employee_avatarUpload employee avatarPOSTAgents: ask firstUploads an avatar image (JPEG, PNG or WebP, at most 5 MB) and returns its url and thumbnail_url, to set as avatar and avatar_thumbnail on a digital employee.

Employee file, 1 operation

get_employee_filesGet employee filesGETAgents: ask firstLists the files of a digital employee's most recent run (files and external_files) and the message they belong to. The docs do not describe a file entry's fields.

Employee goal, 1 operation

update_employee_goalUpdate employee goalPUTAgents: ask firstSets a digital employee's goal: title (required), description, success_criteria and status (planned, active, achieved or cancelled). Send the whole goal: keeping fields left out is not documented.

Employee run output, 1 operation

get_employee_run_outputGet employee run outputGETAgents: ask firstGets the result of a completed digital employee run: output text, rendered output, the input used and the number of tool calls. Not cached.

Employee runs summary, 1 operation

get_employee_runs_summaryGet employee runs summaryGETAgents: ask firstCounts a digital employee's runs (total, completed, failed, running) and gives when the last one ran.

Employee schedule, 1 operation

preview_employee_schedulePreview employee schedulePOSTAgents: ask firstChecks a cron schedule and returns the effective cron and the next planned run times in execution_timezone (UTC by default). Changes nothing: use it before setting heartbeat_cron.

Employee version, 1 operation

list_employee_rollback_optionsList employee rollback optionsGETAgents: ask firstLists the earlier versions of a digital employee you can roll back to (version, changed fields, who changed it, when) and the current version, 5 by default, up to 20.

Workspace usage, 1 operation

list_workspace_usageList workspace usageGETAgents: ask firstLists the workspace's agent executions (who ran which agent, status, credits, model, tokens), last 30 days by default, 20 per page; filter by range, dates, user or type. Includes users' emails.

Use Tess AI from Claude Code, Cursor or Codex

Connect Tess AI in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Tess AI 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 Tess AI

Create your workspace
Sign up free. Nothing to install.
Connect Tess AI once
Add your Tess AI 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.