Skip to content

API reference

The REM Vibe Coder API endpoints for people - chats, messages, runs, plans, questions, projects, code, uploads, memory and billing - with request and response examples.

All paths are under https://remvibecoder.com/api/v1. Authenticate as described in Authentication; in the examples $VC is the base URL and $AUTH stands for the cookie and CSRF headers.

Account

Method and path

What

GET /me

Who you are: id, email, display_name, role

GET /auth/config

(no sign-in) How sign-in works here: remauth_mode (redirect or in_app), links to REM CloudHub people and teams

POST /auth/logout

End the session here and at REM Auth; answers signout_url for the browser

GET /models

The models you may pick: [{"id": "claude-sonnet-5", "label": "Claude Sonnet 5", "tier": "mid"}, …]

GET /stacks

The stacks the engine can pick from (read-only; you never choose one)

Chats and messages

Start a chat with a first message

curl -s -X POST "$VC/chats" $AUTH -H "Content-Type: application/json" -d '{
  "prompt": "Build a habit tracker: add habits with a name and a colour, tick each habit off for today, and see each habit'\''s current streak and the last 7 days.",
  "mode": "auto"
}'
{"chat": {"id": "1b6365e5-afc9-4e3f-9038-5b7e5b2bb8f2", "project_id": null, "title": "Build a habit tracker: add habits with a name and a colou...", "mode": "auto", "archived": false, "last_run_status": "queued"}, "run_id": "bda0f6fc-cb14-4e11-b2f2-d0ed13f805dd"}

Fields of POST /chats: prompt (the first message; empty creates an empty chat), mode (auto or plan), model (an id from GET /models, empty = Auto), project_id (continue an existing project), title.

Send a message

curl -s -X POST "$VC/chats/$CHAT/messages" $AUTH -H "Content-Type: application/json" \
  -d '{"text": "Add a weekly summary screen", "mode": "plan"}'
{"message": {"id": "…", "chat_id": "…", "role": "user", "text": "Add a weekly summary screen", "run_id": "…", "queued": false, "mode": "plan"}, "run_id": "…", "run_status": "queued", "queued": false}

text is at most 20,000 characters. mode: auto (default) or plan. attachments: ids of files uploaded to this chat. When a run is already active the message is queued ("queued": true) and read by that run before its next step. With no credits left the answer is 402 out_of_credits.

Method and path

What

GET /chats

Your chats, newest first, with last_run_id and last_run_status

GET /chats/{id}

One chat

GET /chats/{id}/messages

Its messages (your messages and the assistant's answers)

GET /chats/{id}/plans

Every plan card of the chat: items (text, status, system), approved, database, detail, mode

GET /chats/{id}/turns

How each run was answered: kind (chat, doc, design, build) and whether a sandbox ran

Runs

A run's status is one of queued, running, waiting_approval, done, failed, stopped.

Method and path

Body

What

GET /runs/{id}

status, mode, approved, last_seq, error

GET /runs/{id}/events?after=N

Stored events after sequence N (see Run events)

POST /runs/{id}/stream-ticket

{"ticket", "expires_in", "url"} for an SSE client

GET /runs/{id}/stream?after=N

Server-Sent Events: replay, then live until the run ends

GET /runs/{id}/queue

A queued run's position and reason (your_other_builds, waiting_for_a_worker)

POST /runs/{id}/approve

{"approve": true, "plan": ["…"], "note": ""}

Approve a Plan-mode plan, optionally as edited

POST /runs/{id}/replan

{"instruction": "drop the export", "plan": ["…"]}

A new plan from your instruction

POST /runs/{id}/answer

{"question_id": "…", "answer": "once"}

Answer the open question card: once, always, no, or text (max 2,000 characters)

POST /runs/{id}/stop

Stop after the current step

GET /runs/{id}/usage

Model calls of the run: tokens and cost per model

Example: approve a plan with one line changed.

curl -s -X POST "$VC/runs/$RUN/approve" $AUTH -H "Content-Type: application/json" -d '{
  "approve": true,
  "plan": ["Add a new habit with a name and colour", "Tick off habits completed today", "Weekly summary per habit"]
}'

Projects

Method and path

What

GET /projects

Your projects: name, stack (chosen by the engine), preview_status, live_url, integrations

GET /projects/{id}

One project

PATCH /projects/{id}

{"name": "…", "description": "…"}

DELETE /projects/{id}

Delete the project

GET /projects/{id}/files

The app's file list (exactly what is exported)

GET /projects/{id}/file?path=frontend/src/App.tsx

One file's text

GET /projects/{id}/code.zip

The code as a zip (see Export formats)

GET /projects/{id}/preview

Preview state and address (?wake=true restores a sleeping sandbox)

GET /projects/{id}/tests

The checks of the latest runs with results and screenshots

POST /workspace/projects/{id}/redeploy

Redeploy the latest delivered version

GET /workspace/sidebar

Projects and chats as the sidebar shows them

{"id": "048c3eb3-dfd8-40aa-bdb5-07b1bcce02b0", "owner_id": 3, "name": "Habit tracker with streaks and history", "archived": false, "preview_status": "starting", "live_url": "", "stack": "fastapi-react", "stack_chosen_by": "engine", "integrations": {}, "created_at": "2026-11-18T10:05:32.219Z"}

Uploads

Method and path

What

GET /uploads/policy

max_file_mb, allowed_types, max_files_per_chat

POST /chats/{id}/files

Upload (multipart file); answers the file's id to put in attachments

GET /chats/{id}/files

Files of the chat

GET /files/{id}/download

Download one

Memory

Method and path

What

GET /memory

Your memory entries

POST /memory

Add one

DELETE /memory/{key}

Remove one

Billing

Method and path

What

GET /billing/catalog

(no sign-in) Plans and prices from REM Payments; {"enabled": false} when billing is off

GET /billing/account

Your plan, balance (free, plan, rollover, topup, total, available), low, can_buy

POST /billing/checkout

{"slug": "pro", "return_path": "/billing"} → checkout_url (or an in-app session)

POST /billing/cancel

Cancel at the end of the period

GET /billing/invoices

Your invoices

GET /billing/ledger?limit=50&offset=0

Credit movements

{"enabled": true, "payer": "user:3", "can_buy": true, "plan": {"slug": "free", "code": "free", "name": "Free", "credits_per_cycle": 100}, "status": "free", "balance": {"free": 100, "plan": 0, "rollover": 0, "topup": 0, "total": 100, "available": 100, "held": 0}, "low": true, "credits_name": "Vibe Coder credits"}

Other public endpoints

Method and path

What

GET /health

Service health

GET /showcases/public

The templates: slug, title, category, summary, prompt

POST /feedback

Send feedback (what the account menu's Send feedback uses)

Last updated 18 Nov 2026

API reference | REM Vibe Coder · REM Platforms Blog