circuit-framework

What it is

Circuit Framework is a crypto-native multi-agent research and paper-trading framework where specialized agents analyze market structure, derivatives, sentiment, catalysts and market regime before a deterministic risk engine approves or rejects each trade. It is a fork of TradingAgents (Apache 2.0); user-facing branding and crypto workflows are Circuit Framework. The project emphasizes research only, with no real trades executed and no wallet credentials requested.

How it works

The architecture routes data through a sequence of analytical components: Market Structure, Derivatives, Sentiment, Catalyst, Regime, then Trader, followed by Risk Debate and Portfolio Manager, culminating in a Deterministic Risk Gate that either approves trades or rejects them. All crypto analysts share a single immutable CryptoMarketSnapshot built at the start of the run. Proposals contain entry, stop, take-profits, size/leverage, confidence, thesis, and snapshot_id; risk rules enforce data staleness checks, stop/R:R checks, spread limits, leverage clamps, volatility, and confidence scalars. Paper executions are stored in a Paper Execution SQLite.

Getting started

Key commands from the README:

pip install -e ".[dev]"

tradingagents crypto analyze BTC
tradingagents crypto analyze ETH --strategy momentum
tradingagents crypto analyze SOL --interval 1h
tradingagents crypto analyze HYPE --strategy derivatives --paper
tradingagents crypto portfolio
tradingagents crypto positions
tradingagents crypto leaderboard

Paper DB default: ~/.tradingagents/circuit/paper.db (TRADINGAGENTS_PAPER_DATABASE_PATH).

Programmatic usage example:

from tradingagents.graph.trading_graph import TradingAgentsGraph
from tradingagents.default_config import DEFAULT_CONFIG
from tradingagents.graph.setup import CRYPTO_DEFAULT_ANALYSTS

config = DEFAULT_CONFIG.copy()
ta = TradingAgentsGraph(
    selected_analysts=list(CRYPTO_DEFAULT_ANALYSTS),
    config=config,
    asset_type="crypto",
    strategy_profile="balanced",
)
state, decision = ta.propagate(
    "BTC",
    "2026-07-14",
    asset_type="crypto",
    strategy_profile="balanced",
)

Environment/overrides and testing details are provided, including offline crypto tests and fixtures.

Getting started (installation hints)

Installation notes include:

git clone <this-repo>
cd circuit-framework
python3 -m pip install -e ".[dev]"
cp .env.example .env   # add LLM API keys for live analysis

Docker guidance references a docker compose setup for tradingagents.

Recent releases

There are no listed releases in the repository metadata.

Traction

Stars in the repository: 479. Forks: 19. Open issues: 0. Language: Python. License: Apache-2.0. Created: 2026-07-16. Last push: 2026-07-16.

Behind the repo

The project is a fork of TradingAgents by Tauric Research (Apache 2.0). Upstream attribution is provided with a citation block.

Caveats

• Paper trading and data usage are crypto-focused with Hyperliquid public data; data access is via a public endpoint with no authentication for tests.
• Known limitations include not including on-chain data without a verified provider, potential unavailability of liquidation fields, and that LLM analysts can be wrong; the risk gate is deterministic but cannot invent edge.
• Paper fills use mid ± slippage and configured fees; no real exchange matching.