Continuing conversations
Associate a session with your external_chat_id and send follow-up messages to the same conversation.
RemoteHangar API
Connect an internal portal, help desk or company product to a prepared RemoteHangar workspace. Users stay in your interface; the assistant works with configured sources and tools.
Keep the RemoteHangar token on your application backend. Never expose it in the user’s browser.
Associate a session with your external_chat_id and send follow-up messages to the same conversation.
Receive text, tool activity and completion through SSE. Show execution progress in your own interface.
Select a configured agent and MCP connections instead of rebuilding their setup in each application.
Interrupt requests, close sessions and handle limits and failures through the API contract.
API / HTTP + SSE
Request structures from the existing API. Agent and connection identifiers come from your own installation.
01 / Open a session
POST /v1/sessions
Authorization: Bearer YOUR_REMOTEHANGAR_TOKEN
Content-Type: application/json
{
"external_chat_id": "portal:project-42:conversation-7",
"agent_id": "project-advisor"
}02 / Send a question and read SSE
POST /v1/sessions/SESSION_KEY/prompts
Authorization: Bearer YOUR_REMOTEHANGAR_TOKEN
Content-Type: application/json
Accept: text/event-stream
{
"prompt": "Which tasks are blocking the project launch? Cite your sources."
}Store the session_key returned in the response and use it for follow-up messages. Creating again with the same external_chat_id returns the associated session.
Stream events
The complete OpenAPI schema and interactive documentation are available in your installation when the API is enabled. Available methods and models depend on its configuration.
/v1/agents/v1/mcp-servers/v1/sessions/v1/sessions/:key/prompts/v1/sessions/:key/interrupt/v1/sessions/:keyA RemoteHangar token authorizes your application. It is not a pay-as-you-go model-provider key: execution uses accounts connected in RemoteHangar and their available plans.
Separate RemoteHangar tokens with permitted API operations.
Source access through the assistant’s configured tools.
Limits and errors handled by your application.
Requests and necessary context sent to your chosen model provider.
Choose the interface, questions, sources and permitted actions. The pilot validates the user journey against agreed outcome criteria.
Discuss a pilotWe agree on timing, pricing and scope before the pilot starts.