> ## Documentation Index > Fetch the complete documentation index at: https://docs.opal.dev/llms.txt > Use this file to discover all available pages before exploring further. # Set the agent kill switch > **Prototype (AG-297).** Turns the org-level agent kill switch on or off. Admin only. While it is on, Opal answers every `PreToolUse` hook POST to `/agent-activity` with a `deny` decision, so a cooperating Claude Code client stops running tools. **It is not a security boundary, and it fails open.** It reaches only clients that are configured to POST to `/agent-activity` and that honour the answer; a client on a non-default base URL, an older build, or with no hook deployed is unaffected. If Opal is unreachable or slow the tool call proceeds, because Claude Code treats a failed `PreToolUse` hook as a non-blocking error by design. Activation is always time-bound: `duration_minutes` is capped, and an activation that reaches `expires_at` stops denying on its own. ## OpenAPI ````yaml https://app.opal.dev/openapi.yaml put /agent-kill-switch openapi: 3.1.0 info: contact: email: hello@opal.dev name: Opal Team url: https://www.opal.dev/ description: >- The Opal API is a RESTful API that allows you to interact with the Opal Security platform programmatically. title: Opal API version: '1.0' servers: - description: Production url: https://api.opal.dev/v1 security: [] tags: - name: access-rules description: Operations related to access rules - name: agent-activity description: Operations related to agent activity telemetry - name: apps description: Operations related to apps - name: bundles description: Operations related to bundles - name: campaigns description: Operations related to access review campaigns - name: configuration-templates description: Operations related to configuration templates - name: delegations description: Operations related to request reviewer delegations - name: event-streams description: Operations related to event streaming connections - name: events description: Operations related to events - name: groups description: Operations related to groups - name: group-bindings description: Operations related to group bindings - name: idp-group-mappings description: Operations related to IDP group mappings - name: message-channels description: Operations related to message channels - name: non-human-identities description: Operations related to non-human identities - name: on-call-schedules description: Operations related to on-call schedules - name: opal-queries description: Operations related to OpalQuery - name: opal-scripts description: Operations related to OpalScripts - name: owners description: Operations related to owners - name: requests description: Operations related to requests - name: resources description: Operations related to resources - name: role-assignments description: >- Operations related to role assignments (general grants between an entity and a principal) - name: paladin description: Operations related to Paladin - name: sessions description: Operations related to sessions - name: tags description: Operations related to tags - name: tokens description: Operations related to API tokens - name: uars description: Operations related to UARs. Deprecated in favor of the `campaigns` API. - name: users description: Operations related to users paths: /agent-kill-switch: put: tags: - agent-activity summary: Set the agent kill switch description: >- **Prototype (AG-297).** Turns the org-level agent kill switch on or off. Admin only. While it is on, Opal answers every `PreToolUse` hook POST to `/agent-activity` with a `deny` decision, so a cooperating Claude Code client stops running tools. **It is not a security boundary, and it fails open.** It reaches only clients that are configured to POST to `/agent-activity` and that honour the answer; a client on a non-default base URL, an older build, or with no hook deployed is unaffected. If Opal is unreachable or slow the tool call proceeds, because Claude Code treats a failed `PreToolUse` hook as a non-blocking error by design. Activation is always time-bound: `duration_minutes` is capped, and an activation that reaches `expires_at` stops denying on its own. operationId: setAgentKillSwitch requestBody: content: application/json: schema: $ref: '#/components/schemas/SetAgentKillSwitchRequest' required: true responses: '200': content: application/json: schema: $ref: '#/components/schemas/AgentKillSwitch' description: The state of the kill switch after the change. '400': description: The body was unparseable, or `duration_minutes` was out of range. '403': description: >- The caller is not an admin, or the agent kill switch is not enabled for this organization. security: - BearerAuth: [] components: schemas: SetAgentKillSwitchRequest: description: >- # SetAgentKillSwitchRequest Object ### Description Turns the agent kill switch on or off. Setting `active` to `true` while it is already on supersedes the current activation rather than extending it: the old row keeps its own reason and window as part of the audit trail, and a new one starts. properties: active: description: >- True denies agent tool calls for this organization; false lifts the denial. type: boolean duration_minutes: description: >- How long the denial lasts, from now. Ignored when `active` is false. Capped server-side - a kill switch with no end is how an organization discovers months later that its engineers have been quietly working around it. type: integer reason: description: >- Why. Shown verbatim to the developer whose tool call is denied, so write it for them rather than for the audit log - it is the only explanation they get. type: string required: - active type: object AgentKillSwitch: description: >- # AgentKillSwitch Object ### Description The org-level agent kill switch, and the activation that is in force or last was. `active` is what matters; the rest is the record of how it got that way. An activation is in force only while it has not been deactivated AND `expires_at` is still in the future, and that is evaluated on every read - so `active` can be `false` on a row that was never explicitly turned off. Everything below `active` describes the LATEST activation whether or not it is still in force, so an admin reading an inactive switch can still see who last turned it on and when it ended. properties: active: description: >- Whether agent tool calls are being denied right now. This is the field to read; it already accounts for expiry. type: boolean activated_at: description: >- When the latest activation was turned on. Absent if it never has been. format: date-time type: string activated_by_user_id: description: >- The admin who turned it on. Absent when that user has since been hard-deleted - the activation record outlives them deliberately. format: uuid type: string expires_at: description: >- When the latest activation stops denying on its own. In the past for an activation that already ran out. format: date-time type: string deactivated_at: description: >- When the latest activation was turned off before its expiry, either by an admin or by being superseded. Absent when it ended at `expires_at`, or has not ended. format: date-time type: string reason: description: >- The admin's stated reason. Not an internal note - it is shown to the developer whose tool call is denied, in their own terminal. type: string required: - active type: object securitySchemes: BearerAuth: scheme: bearer type: http ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.