Scaffold member portal: OC webhook + Cloudron user mgmt + LiteLLM key injection

This commit is contained in:
inference-bot committed 2026-09-04 17:06:27 -06:00
commit 451fb561bc
7 files changed
+353

No files matched your search

+5
View File
@@ -0,0 +1,5 @@
__pycache__
*.pyc
.git
*.md
.dockerignore
+19
View File
@@ -0,0 +1,19 @@
{
"id": "coop.inference.member-portal",
"title": "Member Portal",
"author": "Inference Cooperative",
"description": "Membership middleware: syncs Open Collective members with Cloudron users and LiteLLM keys, and injects per-member keys into inference requests.",
"version": "0.1.0",
"healthCheckPath": "/health",
"httpPort": 8000,
"addons": {
"localstorage": {},
"oidc": {
"loginRedirectUri": "/",
"logoutRedirectUri": "/"
}
},
"manifestVersion": 2,
"website": "https://inference.coop",
"contactEmail": "info@inference.coop"
}
+18
View File
@@ -0,0 +1,18 @@
FROM python:3.12-slim
WORKDIR /app/code
# Install dependencies
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
# Copy app
COPY app/ ./app/
# Cloudron runs as UID 1000
RUN mkdir -p /app/data && chown -R 1000:1000 /app/data /app/code
USER 1000
EXPOSE 8000
CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]
+84
View File
@@ -0,0 +1,84 @@
# Member Portal — membership middleware for the Inference Cooperative
Reconciles three systems into one membership lifecycle:
| System | Role | Source of truth for |
|--------|------|---------------------|
| Open Collective | Billing | Who is a *paying* member |
| Cloudron | Identity/SSO | Who can *log in* |
| LiteLLM | Inference | Who can *use models*, and how much |
## Architecture
```
Open Collective (billing)
│ webhook
▼
┌─────────────────────────────┐
│ Member Portal (this app) │
│ - OC webhook handler │
│ - Cloudron user management │
│ - LiteLLM key management │
│ - key-injection proxy │
└─────────────────────────────┘
▲
│ X-User-Email header (LibreChat forwards identity)
│
LibreChat ──→ Member Portal ──→ LiteLLM ──→ backend
```
## The three responsibilities
1. **Webhook handler** (`POST /webhook/opencollective`) — listens for Open
Collective membership events and activates/deactivates members.
2. **Key injector** (`/v1/*`) — reads the member's email from the
`X-User-Email` header, looks up their LiteLLM key, and forwards the
request to the gateway with that key.
3. **Admin** (`/health`, `/`) — health and status.
## Configuration (environment variables)
| Variable | Purpose |
|----------|---------|
| `CLOUDRON_API` | Cloudron API origin (auto-set by Cloudron) |
| `CLOUDRON_TOKEN` | Cloudron API token (Read+Write) |
| `LITELLM_BASE` | LiteLLM gateway URL |
| `LITELLM_MASTER_KEY` | LiteLLM master key |
| `OPENCOLLECTIVE_WEBHOOK_SECRET` | Optional webhook signature secret |
## Tier budgets
Governance decision (set in Loomio). Defaults:
- `free` — $2/month of tokens
- `member` — $15/month
- `supporter` — $30/month
## LibreChat wiring
Point LibreChat's custom endpoint at this portal (not directly at LiteLLM),
and add the identity header:
```yaml
endpoints:
custom:
- name: "Inference Cooperative"
apiKey: "${LITELLM_KEY}"
baseURL: "https://portal.inference.coop/v1"
headers:
X-User-Email: "{{LIBRECHAT_USER_EMAIL}}"
models:
default: ["deepseek-v4-flash", "gpt-oss-120b"]
fetch: true
```
## Status
**Scaffold** — the three handlers are stubbed with the correct integration
points. Remaining work before production:
- [ ] Verify Open Collective webhook event names + payload shape
- [ ] Verify Cloudron user-creation API payload (role/group assignment)
- [ ] Add a persistent store (SQLite/Postgres) for email→key mapping
- [ ] Add HMAC signature verification for the OC webhook
- [ ] Wire the OIDC addon for admin access
+3
View File
@@ -0,0 +1,3 @@
from .main import app
__all__ = ["app"]
+220
View File
@@ -0,0 +1,220 @@
"""
Member Portal — membership middleware for the Inference Cooperative.
Reconciles three systems:
1. Open Collective — who is a paying member (billing)
2. Cloudron — who can log in (identity/SSO)
3. LiteLLM — who can use the models, and how much (inference)
Three responsibilities:
A. Webhook handler — listen for Open Collective membership events
B. Key injector — read the member's email from a header, inject their
LiteLLM key, and forward the request to the gateway
C. Admin endpoints — health, status, manual reconciliation
"""
import os
import json
import logging
import httpx
from fastapi import FastAPI, Request, Response, HTTPException
from fastapi.responses import JSONResponse
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger("member-portal")
app = FastAPI(title="Inference Cooperative Member Portal")
# --- Configuration (from environment) ---
CLOUDRON_API = os.environ.get("CLOUDRON_API_ORIGIN", "https://my.inference.coop")
CLOUDRON_TOKEN = os.environ.get("CLOUDRON_TOKEN", "")
LITELLM_BASE = os.environ.get("LITELLM_BASE", "https://gateway.inference.coop")
LITELLM_MASTER_KEY = os.environ.get("LITELLM_MASTER_KEY", "")
OPENCOLLECTIVE_SECRET = os.environ.get("OPENCOLLECTIVE_WEBHOOK_SECRET", "")
# Tier → monthly budget (in USD of tokens). Governance decision, set in Loomio.
TIER_BUDGETS = {
"free": 2.0,
"member": 15.0,
"supporter": 30.0,
}
DEFAULT_TIER = "member"
# --- Helpers ---
def cloudron_headers() -> dict:
return {"Authorization": f"Bearer {CLOUDRON_TOKEN}"}
async def cloudron_create_user(email: str, name: str) -> str:
"""Create (or return existing) Cloudron user, assigned to the chat app."""
async with httpx.AsyncClient() as client:
# Check if user exists
r = await client.get(
f"{CLOUDRON_API}/api/v1/users",
headers=cloudron_headers(),
params={"email": email},
)
r.raise_for_status()
users = r.json().get("users", [])
if users:
return users[0]["id"]
# Create user
r = await client.post(
f"{CLOUDRON_API}/api/v1/users",
headers=cloudron_headers(),
json={"email": email, "displayName": name, "role": "user"},
)
r.raise_for_status()
return r.json()["id"]
async def cloudron_set_active(user_id: str, active: bool) -> None:
async with httpx.AsyncClient() as client:
r = await client.post(
f"{CLOUDRON_API}/api/v1/users/{user_id}",
headers=cloudron_headers(),
json={"active": active},
)
r.raise_for_status()
async def litellm_create_key(email: str, budget: float) -> str:
"""Create a LiteLLM virtual key for a member with a budget cap."""
async with httpx.AsyncClient() as client:
r = await client.post(
f"{LITELLM_BASE}/key/generate",
headers={"Authorization": f"Bearer {LITELLM_MASTER_KEY}"},
json={
"key_alias": f"member:{email}",
"max_budget": budget,
"budget_duration": "30d",
"models": ["deepseek-v4-flash", "gpt-oss-120b"],
},
)
r.raise_for_status()
return r.json().get("key", "")
async def litellm_disable_key(email: str) -> None:
"""Disable a member's key (on payment lapse)."""
async with httpx.AsyncClient() as client:
# Find the key by alias
r = await client.get(
f"{LITELLM_BASE}/key/list",
headers={"Authorization": f"Bearer {LITELLM_MASTER_KEY}"},
)
r.raise_for_status()
for k in r.json().get("keys", []):
if k.get("key_alias") == f"member:{email}":
await client.post(
f"{LITELLM_BASE}/key/delete",
headers={"Authorization": f"Bearer {LITELLM_MASTER_KEY}"},
json={"keys": [k["token"]]},
)
return
# --- A. Open Collective webhook ---
@app.post("/webhook/opencollective")
async def opencollective_webhook(request: Request):
"""Handle Open Collective membership events."""
payload = await request.json()
# Verify webhook secret if configured
if OPENCOLLECTIVE_SECRET:
sig = request.headers.get("x-oc-signature", "")
# TODO: verify HMAC signature
if not sig:
raise HTTPException(401, "Missing signature")
event_type = payload.get("type", "")
data = payload.get("data", {})
member = data.get("member", {}) or data.get("fromCollective", {})
email = member.get("email") or data.get("email")
name = member.get("name", "Member")
if not email:
logger.warning("Webhook without email: %s", event_type)
return JSONResponse({"status": "ignored", "reason": "no email"})
logger.info("Open Collective event: %s for %s", event_type, email)
if event_type in ("collective.member.created", "collective.transaction.created"):
# New or renewed member → ensure active
tier = (data.get("tier") or {}).get("slug", DEFAULT_TIER)
budget = TIER_BUDGETS.get(tier, TIER_BUDGETS[DEFAULT_TIER])
user_id = await cloudron_create_user(email, name)
await cloudron_set_active(user_id, True)
await litellm_create_key(email, budget)
return JSONResponse({"status": "activated", "user_id": user_id, "budget": budget})
if event_type in ("collective.member.deleted", "collective.transaction.deleted"):
# Lapsed member → deactivate
await litellm_disable_key(email)
return JSONResponse({"status": "deactivated"})
return JSONResponse({"status": "ignored", "type": event_type})
# --- B. Key injector (proxy) ---
@app.api_route("/v1/{path:path}", methods=["GET", "POST", "PUT", "DELETE", "PATCH"])
async def inject_key(request: Request, path: str):
"""Read the member's email from a header, inject their key, forward to LiteLLM."""
email = request.headers.get("x-user-email", "")
if not email:
raise HTTPException(401, "No member identity (x-user-email header)")
# Look up the member's key (in production: from a store keyed by email)
# For MVP: derive deterministically or look up via LiteLLM
async with httpx.AsyncClient() as client:
r = await client.get(
f"{LITELLM_BASE}/key/list",
headers={"Authorization": f"Bearer {LITELLM_MASTER_KEY}"},
)
r.raise_for_status()
member_key = None
for k in r.json().get("keys", []):
if k.get("key_alias") == f"member:{email}":
member_key = k.get("token")
break
if not member_key:
raise HTTPException(403, "No active membership key")
# Forward the request to LiteLLM with the member's key
body = await request.body()
headers = dict(request.headers)
headers["authorization"] = f"Bearer {member_key}"
headers.pop("host", None)
headers.pop("content-length", None)
async with httpx.AsyncClient() as client:
upstream = await client.request(
method=request.method,
url=f"{LITELLM_BASE}/v1/{path}",
headers=headers,
content=body,
)
return Response(
content=upstream.content,
status_code=upstream.status_code,
headers={"content-type": upstream.headers.get("content-type", "application/json")},
)
# --- C. Admin / health ---
@app.get("/health")
async def health():
return {"status": "ok"}
@app.get("/")
async def index():
return {"service": "Inference Cooperative Member Portal", "version": "0.1.0"}
+4
View File
@@ -0,0 +1,4 @@
fastapi==0.115.0
uvicorn[standard]==0.30.6
httpx==0.27.2
pydantic==2.9.2