What is on in your workspace
Which nima capabilities are on for everyone, which an owner or admin must turn on, which the nima team enables on request, and how to turn each one on.
Every nima capability is in one of four states for your workspace: on for every workspace, off until an owner or admin turns it on, enabled by the nima team on request, or not available yet. This page is the one place that lists which is which and how to change it. Other pages link here instead of repeating it.
Availability as of 2026-09-30. None of the switches in the second group has a control in the admin panel. The workspace settings change only with PATCH /organizations/settings from a signed-in owner or admin session. The writer and hosted-compute opt-ins can also be set over MCP with an integrations key or with a /nima command in Slack. Or ask the nima team.
Capabilities on for every workspace
These run for every workspace with no setting to change. Some need an integration connected first, as noted.
| Capability | What you get | What it needs |
|---|---|---|
| Ingestion from integrations | Signals from Slack, GitHub, Linear, PostHog, Sentry, ClickUp, Granola and webhooks. Jira issues sync to cards, not signals | The integration connected on the Integrations screen. A webhook endpoint is created with POST /v1/signal-webhooks from a signed-in owner or admin session, or ask the nima team (see webhooks) |
| Automatic beacon placement | nima places each new signal in the open beacon it belongs to, or starts a new beacon for it when none fits | Nothing. See how nima decides |
| Slack answers | Mention @nima or DM it and it answers from your signals, specs and code | Slack connected |
| Scheduled messages and reminders | One-time reminders to your own Slack DM, set from Slack or MCP | Your Slack account linked to your nima account |
| Knowledge graph | Organization memory that answers who asked for something and why a decision was made. See the knowledge graph | Nothing |
| Code tools | Code search and blast radius over indexed repositories. See code intelligence | A repository indexed through the GitHub integration |
| MCP server | The tools at https://mcp.getnima.io/mcp. See connect to nima's MCP server | An OAuth sign-in or an API key |
| REST API | The routes in the API reference | An API key or OAuth token for most routes; workspace settings need a signed-in session |
Two pipeline behaviors are not in this group. Auto-promote of fix beacons and the daily beacon sweep are off for every workspace; the nima team controls them platform-wide (see the section on team-enabled capabilities below).
Capabilities an owner or admin turns on
An owner or admin of the workspace turns these on. Each one is off by default for every workspace, even though the feature itself is live on the platform.
| Capability | Setting | Default | What it does |
|---|---|---|---|
| Watches and triggers | triggersEnabled | off | Lets events recorded for your workspace fire watches and reach people. See watches and reminders |
| Outbound actions | dispatchEnabled | off | Lets nima write into integrations for your workspace: Slack messages from rules, ticket comments, questions to people. See outbound actions |
| Real @mentions in automated messages | dispatch.mentionsEnabled | off | With it off, automated messages show plain display names instead of @mentions |
| Internal-only tracker projects | dispatch.internalProjects | empty list | Marks tracker projects that have no external readers. Until a project is listed, nima treats every tracker project as possibly readable by outsiders. You can list up to 100 projects, each as a provider and a projectKey |
| Daily cost cap on agent-handler runs | dispatch.dailyUsdCap | 5 (US dollars, per UTC day) | A number from 0 to 1000. A run over the cap is refused. Set 0 to stop every agent-handler run |
Outbound actions also depend on the integrations you connect. A watch or rule only reaches a channel that is connected, and ticket replies need a writer credential (next table). Turning on dispatchEnabled does not by itself let nima post anywhere.
Rules that send Slack messages or ticket replies run on the watch engine, so they need triggersEnabled as well as dispatchEnabled; creating a rule with either off returns 409. Questions to people and approvals do not need triggersEnabled.
Writers and opt-ins that have their own switch
These are separate from the workspace settings. Each is off until an owner or admin connects or enables it, either over MCP with a key that holds the integrations:write grant (an owner or admin creates one in Slack with /nima mcp-key integrations) or over REST with that key or a signed-in owner or admin session. The routes are listed in writers.
| Capability | How an admin turns it on |
|---|---|
| Hosted compute for agent operations | After the nima team sets your workspace up (see the team-enabled section), an admin sets enabled to true with the update_agent_compute_settings tool or PATCH /v1/agent-compute. See hosted VMs and agent compute |
| GitHub publishing | Get the install link with get_github_writer_install_url (or GET /v1/github-writer/app/install), install the optional GitHub writer App, then call set_github_publishing with enabled: true. Installing the App does not enable publishing. An owner or admin can also run /nima github-writer install and /nima github-writer enable in Slack. See GitHub |
| Jira comments | Connect a service account with POST /v1/integrations/jira/writer (email plus a scoped API token), or reuse the person who connected Jira with connect_jira_writer, mode: "connector_fallback" and acknowledged: true. See Jira |
| Linear comments | Not available yet. You can already enable the Linear writer with connect_linear_writer and acknowledged: true, but nothing proposes a Linear reply today. See Linear |
Turning on hosted compute does not grant publishing to GitHub, and turning on GitHub publishing does not turn on compute. The two consents are independent.
Turning on workspace settings with PATCH /organizations/settings
The workspace settings in the table above (triggersEnabled, dispatchEnabled and the dispatch fields) are changed only with PATCH /organizations/settings, called with a signed-in owner or admin session. The admin panel has no control for them, no API key can write them, and there is no MCP tool for them. dispatch.internalProjects can also be changed one project at a time with the assert_project_audience tool or POST /v1/dispatch/project-audience, which need the integrations:write grant (see outbound MCP tools). If you cannot make these calls, ask the nima team and they will do it for you.
A minimal body that turns on watches:
{ "triggersEnabled": true }To turn on outbound actions and set the daily cap in one call:
{
"dispatchEnabled": true,
"dispatch": { "mentionsEnabled": false, "dailyUsdCap": 5 }
}The call merges what you send into the stored settings and returns the full settings object, so you can confirm the result from the response. A malformed dispatch value is rejected before anything is written. Changes to the dispatch object are recorded in an audit trail. dispatch.internalProjects is the exception: it replaces the whole list, and [] clears it. Members without the owner or admin role get a 403.
The settings route, its fields and the full response are documented in workspace settings. Roles are covered in members and roles.
The Signals per beacon setting does not change grouping
Settings, then Grouping shows a "Signals per beacon" value (beaconMinClusterSize, 2 or 3). nima stores it but grouping no longer reads it, so changing it has no effect on how beacons form. Do not use it to tune grouping.
Capabilities the nima team enables on request
The nima team, not your admins, controls these. Ask the team to turn one on.
| Capability | State | Notes |
|---|---|---|
| Hosted VMs for agent operations | Enabled by the nima team on request | The team grants your plan the hosted-compute feature and sets operator pricing. Then an admin opts in with the compute setting above. Operations that run on your own harness do not need any of this. See agent operations |
| HubSpot ingestion | Rolling out, ask the nima team | Connecting HubSpot, scanning it and saving a field-mapping recipe work today. The import does not queue: POST /v1/hubspot/import answers queued: false, and HubSpot signals do not arrive yet. See HubSpot |
| Auto-promote of fix beacons | Not available yet | Off for every workspace. This is one platform-wide switch that the nima team controls, so it cannot be turned on for a single workspace. Today a spec starts only when a person promotes a beacon |
| Daily beacon sweep and automatic parking of signals nobody can build from | Not available yet | Off for every workspace. This is one platform-wide switch that the nima team controls, so it cannot be turned on for a single workspace. If it is turned on, nima only holds a beacon and gives a reason; a person confirms any merge or dismissal |
| Copilot fact capture from chat | Not available yet | Off for every workspace. The chatFactCaptureEnabled setting exists in the settings object but has no effect while the capability is off |
Plan features that can gate a route
Some routes are declared under a plan feature, and a workspace whose plan lacks the feature can get a 403 INSUFFICIENT_PERMISSIONS with details.feature set to the feature name. Handle that response, and ask the nima team to change the plan if you hit it.
The features are slack_ingestion, github_ingestion, copilot_chat, boards, documents, collaboration, code_intelligence, prd_generation and agent_compute. New workspaces start on the Free plan. Do not assume the Free plan blocks a route because it lacks the feature: whether a call is stopped depends on the route, so read the response, not the plan table.
The agent_compute feature is different. It is checked when a hosted operation starts or when an admin enables hosted compute, it is off on every standard plan, and only the nima team can grant it. Nobody is billed today; billing is not enabled.
See errors for the response shape.
Checking what is on
You can read the current state of each switch without changing anything.
| To see | Call | Who can call it |
|---|---|---|
triggersEnabled, dispatchEnabled, the dispatch object and the other workspace settings | GET /organizations/settings | Any member, with a signed-in session |
| Hosted compute opt-in, budgets and limits | get_agent_compute_settings tool, or GET /v1/agent-compute | An owner or admin: signed-in session, or a key with integrations:write |
| GitHub writer connection and publishing consent | get_github_publishing_status tool, or GET /v1/github-writer/app/status | An owner or admin: signed-in session, or a key with integrations:write |
| Jira and Linear writer status | GET /v1/integrations/jira/writer, GET /v1/integrations/linear/writer | An owner or admin |
GET /organizations/settings is a session route. It does not accept an API key. A 403 from any route on this page means the caller lacks the owner or admin role or the grant; members and roles says which.
Members, roles and workspaces
Owner, admin and member permissions, invitations, removing people, owning several workspaces, and deleting a workspace.
Connect to nima's MCP server
How to connect an MCP client to https://mcp.getnima.io/mcp with OAuth or an API key, which permissions it asks for, which tools you will see, and how to fix a missing tool.