RugMunchMedia/rmi-backend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

docs: rewrite README and add PRODUCT, ARCHITECTURE, ROADMAP, TODO per governance framework

Browse source
This commit is contained in:
Crypto Rug Munch 2026-07-03 15:56:26 +02:00
parent b2cdfce4cd
commit bbb9a9291d
5 changed files with 212 additions and 87 deletions
Download patch file Download diff file Expand all files Collapse all files

61
ARCHITECTURE.md Normal file
View file

@ -0,0 +1,61 @@
# RMI Backend — Architecture
**Status:** canonical
**Last updated:** 2026-07-03
## Overview
FastAPI monolith serving crypto risk intelligence via HTTP, Telegram, and MCP.
## High-level flow
```
Client (Web / Telegram / AI Agent)
↓
nginx → rmi-backend container (:8000)
↓
FastAPI routers
↓
Services / scanners / DataBus providers
↓
Postgres / Redis / Qdrant / Neo4j / ClickHouse
```
## Components
| Path | Purpose |
|---|---|
| `app/routers/` | HTTP route handlers (thin) |
| `app/scanners/` | Token risk scanners |
| `app/databus/` | 30+ third-party data providers |
| `app/telegram_bot/` | Telegram bot command handlers |
| `app/mcp/` | MCP manifest and tool manager |
| `app/facilitators/` | x402 payment verification |
| `app/wallet_memory/` | Wallet label + clustering subsystem |
## Storage
| Store | Use | Container / Host |
|---|---|---|
| Postgres | Primary OLTP, Ponder indexer data | `rmi-postgres` |
| Redis | Cache, Celery queue | `rmi-redis` |
| Qdrant | Vector embeddings | `rmi-qdrant` (down) |
| Neo4j | Wallet graph / labels | `rmi-neo4j` |
| ClickHouse | Analytics, 39M address labels | `rmi-clickhouse` |
## Interfaces
| Endpoint | Purpose |
|---|---|
| `GET /health` | Health check (currently unhealthy due to Redis) |
| `GET /docs` | Swagger/OpenAPI docs |
| `POST /scanner/scan` | Live token scan |
| `POST /mcp/...` | MCP tool endpoints |
## Current issues
- `x402_tools.py` (212 KB) and `x402_enforcement.py` (108 KB) violate the 500-line limit and should become separate services or split by domain.
- `sqlite3` is used in 5 files for local state; should migrate to Postgres/aiosqlite.
- F-string logs and `print()` statements prevent structured log aggregation.
## Deployment
```bash
ssh netcup
cd /root/backend/
docker compose pull
docker compose up -d rmi-backend
```

54
PRODUCT.md Normal file
View file

@ -0,0 +1,54 @@
# RMI Backend — Product Charter
**Product:** Rug Munch Intelligence (RMI) Backend
**Repo:** `git.rugmunch.io/RugMunchMedia/rmi-backend`
**Deploy path:** `/root/backend/` on Talos
**Container:** `rmi-backend`
**Health:** `http://127.0.0.1:8000/health`
**Status:** Production / degraded
## Problem
Crypto users and platforms need a single, fast, machine-readable source of token risk, wallet labels, and on-chain intelligence. Existing tools are fragmented, expensive, or closed.
## User
- Telegram users via `@rugmunchbot`
- RMI web frontend at `https://rugmunch.io`
- AI agents via MCP
- API consumers via x402 pay-per-call
## Scope
- FastAPI backend with token scanner, wallet forensics, market data, alerts
- DataBus: unified data plane across 30+ providers
- Telegram bot commands: `/scan`, `/portfolio`, `/simulate`, `/chart`, `/watch`
- MCP server for agent tool calling
- x402 payment gating for premium endpoints
## Anti-scope
- No frontend code (that lives in `rmi-frontend`)
- No wallet generation (that lives in `walletpress`)
- No web scraping engine (that lives in `pryscraper`)
- No social protocol clients (that lives in `degenfeed-web`)
## Success criteria
- `/health` returns 200
- CI gates are authoritative
- Test coverage >60%
- All MCP tools have Pydantic input/output schemas
- No production file >500 lines
## Current reality
- 990 routes across modular routers
- Health is **unhealthy** because `rmi-redis:6379` does not resolve
- 8.61% test coverage
- 17 `time.sleep()`, 1,713 f-string logs, 399 `print()` calls
- Largest file `app/routers/x402_tools.py` is 212 KB
## Fleet dependencies
- Postgres `rmi-postgres:5432`
- Redis `rmi-redis:6379` (currently failing)
- Qdrant `127.0.0.1:6333` (currently down due to stale IP bind)
- Neo4j `127.0.0.1:7687`
- ClickHouse `127.0.0.1:8123`
## Standards
This repo follows [Rug Munch Media standards](https://git.rugmunch.io/RugMunchMedia/standards).

130
README.md
View file

@ -1,97 +1,53 @@
# RMI — Rug Munch Intelligence
# RMI Backend
> **The first open-source crypto intelligence platform.**
> Bloomberg-grade security and market intelligence for the on-chain world.
**Repo:** `git.rugmunch.io/RugMunchMedia/rmi-backend`
**Deploy:** `/root/backend/` on Talos
**Container:** `rmi-backend`
**Health:** `http://127.0.0.1:8000/health` (currently unhealthy — Redis host not resolving)
---
## What it is
FastAPI backend for the RMI platform. Serves token risk scans, wallet forensics, market data, Telegram bot commands, MCP agent tools, and x402-gated premium endpoints.
## About Rug Munch Media LLC
## Status
Production, but degraded. The container runs and responds, but `/health` reports unhealthy because the Redis hostname `rmi-redis:6379` is not resolving inside the container network.
We are **Rug Munch Media LLC**, a Wyoming-based software company building open-source tools that make the crypto economy safer — for consumers, builders, and institutions.
Crypto moves fast, and bad actors move faster. Phishing sites clone trusted brands within hours of a certificate being issued. Token launches rug in seconds. Wallets drain before a user knows what happened. The existing security stack is fragmented, expensive, and gated behind paywalls. **We are building the open alternative.**
RMI (Rug Munch Intelligence) is the **first open-source crypto intelligence platform** — a unified, AI-powered command center for token risk analysis, wallet forensics, multi-chain market data, phishing detection, and regulatory-grade intelligence. Our ambition is to become the **Bloomberg terminal of crypto**: one trusted surface where the entire industry can see what is happening, what is risky, and what is real.
**Who we serve:**
- **Consumers** — real-time protection against rug pulls, phishing, and wallet drainers
- **Industry** — exchanges, custodians, compliance teams, and researchers who need defensible on-chain intelligence
- **AI agents** — first-class Model Context Protocol (MCP) integration and x402 micropayments for autonomous tooling
We are an open-source company. Our code, our data, and our threat intelligence are public. Our business model is delivering the managed product, the API, and the institutional tier — not locking away the safety net.
---
## What RMI does
### Threat detection
- **Token scanning** — real-time rug-pull, honeypot, and scam detection across 18+ EVM chains + Solana
- **Wallet forensics** — track wallet behavior and transaction patterns, flag known bad actors
- **Phishing detection** — Certificate Transparency monitoring catches phishing domains within hours of registration, before they go live
- **Bayesian reputation** — statistically grounded deployer trust scoring, not vibes
### Market intelligence
- **2,500+ assets** — full coverage across major chains
- **News aggregation** — 1,800+ sources with clustering, dedup, and sentiment
- **Whale alerts** — large wallet movements and trading patterns in real time
- **Multi-chain market data** — OHLCV, liquidity, and volume from a single API
### AI-native platform
- **RAG-grounded reports** — every claim cited, no hallucination
- **MCP server** — drop RMI into any AI agent or LLM workflow
- **x402 micropayments** — pay-per-call pricing for autonomous agents
- **Open weights and open data** — no vendor lock-in
---
## Architecture
RMI is a single FastAPI backend with modular domain services:
```
Backend (FastAPI + Python)
├── DataBus — single data plane, 18 chains, 4-layer defense (labels → scanner → RAG → prices)
├── SENTINEL — multi-chain token security scanner
├── Wallet Bank — 39M+ labeled wallets, federated from open sources
├── RAG Engine — 3-pillar hybrid retrieval (dense + sparse + entity)
├── x402 Gateway — AI-agent micropayments
├── MCP Server — 221 tools for AI agents
└── Observability — Prometheus, GlitchTip, OpenTelemetry
## Quick start (local)
```bash
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python -m pytest tests/unit/ -x -q
```
Storage: PostgreSQL, Redis, Neo4j, Qdrant, ClickHouse, DuckDB.
## Deploy
```bash
ssh netcup
cd /root/backend/
docker compose pull
docker compose up -d rmi-backend
```
---
## Fleet dependencies
| Service | Address | Status |
|---|---|---|
| Postgres | `rmi-postgres:5432` | Up |
| Redis | `rmi-redis:6379` | Failing to resolve |
| Qdrant | `127.0.0.1:6333` | Down (stale IP bind) |
| Neo4j | `127.0.0.1:7687` | Up |
| ClickHouse | `127.0.0.1:8123` | Up |
## Repositories
## Current issues
- 8.61% test coverage
- 17 `time.sleep()`, 1,713 f-string logs, 399 `print()` calls
- Largest file `app/routers/x402_tools.py` is 212 KB
- CI gates use `continue-on-error: true` and `|| true`
- 5 Telegram tokens leaked in git history
| Platform | URL |
|----------|-----|
| 🐙 GitHub (canonical) | https://github.com/Rug-Munch-Media-LLC/rugmuncher-backend |
| 🦊 GitLab (mirror) | https://gitlab.com/cryptorugmuncher/rugmuncher-backend |
| 🤗 HuggingFace (canonical) | https://huggingface.co/cryptorugmuncher/rugmuncher-backend |
## Docs
- `PRODUCT.md` — charter and scope
- `ARCHITECTURE.md` — components and data flow
- `ROADMAP.md` — next 90 days
- `TODO.md` — current tasks
---
## Links
- 🌐 **Website**: https://rugmunch.io
- 📧 **Contact**: info@rugmunch.io
- 💬 **Telegram**: https://t.me/cryptorugmuncher
- 🐦 **X / Twitter**: https://x.com/cryptorugmunch
- 🐙 **GitHub org**: https://github.com/Rug-Munch-Media-LLC
- 🦊 **GitLab**: https://gitlab.com/cryptorugmuncher
- 🤗 **HuggingFace org**: https://huggingface.co/cryptorugmuncher
---
## License
**MIT** — see `LICENSE.md`.
RMI is open source because the safety of the crypto economy cannot depend on a single vendor. We compete on the quality of the product, not on who is allowed to see the threat.
---
**Built by Rug Munch Media LLC · Wyoming, USA**
*Open-source crypto intelligence. AI-native. Built to last.*
## Standards
This repo follows [Rug Munch Media standards](https://git.rugmunch.io/RugMunchMedia/standards).

33
ROADMAP.md Normal file
View file

@ -0,0 +1,33 @@
# RMI Backend — Roadmap
**Next 90 days only.** See `TODO.md` for current session tasks.
## Week 1 — Stabilize
- Fix Redis host resolution for `rmi-backend` health
- Fix Qdrant stale Tailscale IP bind
- Rotate leaked secrets (Telegram tokens, API keys)
- Harden public service binds
## Week 2 — Gates
- Make CI authoritative (remove `|| true` and `continue-on-error`)
- Add `web3-safety` and `hallucination-check` pre-commit hooks
- Get `ruff check` and `ruff format --check` green
## Week 3-4 — Standards Autofix
- Replace all `time.sleep()` with `asyncio.sleep()`
- Convert f-string logs to structured logs
- Remove/replace `print()` statements
- Migrate `sqlite3` to async Postgres
- Add `# noqa: ANY` or replace `Any` annotations
## Week 5-6 — Modularize
- Split `x402_tools.py` into domain modules
- Split `x402_enforcement.py`
- Split `databus/providers.py`
- Begin extracting x402/MCP into standalone product boundaries
## Acceptance
- `/health` green
- CI fails on lint/secrets
- Coverage >60%
- No file >500 lines

21
TODO.md Normal file
View file

@ -0,0 +1,21 @@
# RMI Backend — TODO
## Active
- [ ] Fix `rmi-redis:6379` name resolution so `/health` returns healthy
- [ ] Fix Qdrant stale IP bind (`100.100.18.18:6333` → `127.0.0.1:6333`)
- [ ] Rotate 5 Telegram tokens leaked in `data/cryptoscamdb_urls.yaml` history
- [ ] Rotate Solana Tracker, Arkham WS, Stripe, Helius webhook secrets
- [ ] Remove `continue-on-error: true` and `|| true` from `.github/workflows/ci.yml`
## Next
- [ ] Replace 17 `time.sleep()` calls with `asyncio.sleep()`
- [ ] Convert top 10 f-string log files to structured logging
- [ ] Delete or migrate 5 `sqlite3` usages
- [ ] Split `app/routers/x402_tools.py` (>4,000 lines) into modules
## Blocked
- GitHub external mirror pushes until org account flag is lifted
## Done
- HuggingFace backups consolidated to single bucket
- External mirrors active on GitLab + Codeberg