Juniper is an autonomous agent that watches pump.fun in real time, tracking token volume, bonding-curve migrations (the jump to PumpSwap), and surfacing statistically favorable windows to trade.
🛠️ Built with grok.me — this repo is the scaffold: structure, docs, and interfaces. The hardcore signal logic, model tuning, and live execution engine are being built and iterated on inside grok.me. Think of this repo as Juniper's skeleton; grok.me is where the muscle gets attached.
Pump.fun is noisy — hundreds of tokens launch daily, most die in the first hour, and the few that "migrate" to PumpSwap tend to do so after a recognizable volume ramp. Juniper exists to cut through that noise:
| Module | Job |
|---|---|
juniper_volume_pulse.py |
Streams real-time trade volume per token, computes rolling velocity (SOL/min), and flags abnormal spikes |
juniper_migration_radar.py |
Watches bonding-curve completion %, predicts time-to-migration, and fires alerts when a token crosses the PumpSwap migration threshold |
juniper_timing_oracle.py |
Aggregates historical volume/migration data into a heatmap of best trading windows (hour-of-day, day-of-week) |
juniper_agent.py |
The orchestrator — wires the three modules together into a single polling/streaming loop and dispatches alerts |
juniper_config.py |
Central config: RPC endpoints, polling intervals, thresholds, alert channels |
┌───────────────────────┐
│ pump.fun API │
│ + Solana RPC/WS feed │
└───────────┬─────────────┘
│
┌───────────▼─────────────┐
│ juniper_agent.py │
│ (orchestration loop) │
└───┬───────────┬──────────┘
┌──────────────┘ └──────────────┐
▼ ▼
┌─────────────────────────┐ ┌─────────────────────────────┐
│ juniper_volume_pulse.py │ │ juniper_migration_radar.py │
│ - rolling volume windows │ │ - bonding curve % tracking │
│ - spike detection │ │ - migration ETA estimation │
└────────────┬─────────────┘ └───────────────┬───────────────┘
│ │
└─────────────────────┬───────────────────────┘
▼
┌───────────────────────────┐
│ juniper_timing_oracle.py │
│ - historical aggregation │
│ - best-hour / best-day heat │
└───────────────┬───────────────┘
▼
alerts / logs / dashboard
juniper/
├── README.md ← you are here
├── GROKME_BUILD_NOTES.md ← what's scaffolded here vs. built in grok.me
├── juniper_agent.py ← main entrypoint / orchestrator
├── src/
│ ├── juniper_volume_pulse.py ← volume tracking engine
│ ├── juniper_migration_radar.py ← migration detection engine
│ ├── juniper_timing_oracle.py ← best-time-to-trade analytics
│ ├── juniper_config.py ← config + constants
│ └── juniper_alerts.py ← alert dispatch (console/webhook/discord)
├── data/
│ └── .gitkeep
├── .github/workflows/juniper_ci.yml
├── requirements.txt
├── .env.example
├── .gitignore
└── LICENSE
git clone https://github.com/YOUR_USERNAME/juniper.git
cd juniper
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # already pre-filled with sane public defaults
python juniper_agent.pyJuniper will start polling pump.fun for new token launches, streaming volume data, and logging migration candidates to data/juniper_migrations.log.
All tunables live in src/juniper_config.py:
POLL_INTERVAL_SECONDS— how often Juniper checks for new dataVOLUME_SPIKE_MULTIPLIER— how far above rolling average counts as a "spike"MIGRATION_THRESHOLD_PCT— bonding curve completion % that triggers a migration alertTIMING_WINDOW_DAYS— lookback window (in days) for the best-time-to-trade heatmap
- Repo scaffold + module boundaries (this repo)
- Live pump.fun WebSocket ingestion (in progress on grok.me)
- Migration prediction model v1 (in progress on grok.me)
- Best-time-to-trade heatmap dashboard
- Discord/Telegram alert bot
- Backtesting harness against historical migrations
Juniper is a research/tooling project for tracking public on-chain data. It is not financial advice, does not execute trades on its own, and pump.fun tokens are extremely high-risk, high-volatility assets. Nothing here guarantees profitability. DYOR.
MIT — see LICENSE.
🌿 Juniper — scaffolded here, sharpened on grok.me.