HyperTrade is a crypto trading agent for market research and execution. Current V1 covers observable Agent graph runs, provider routing, tool calling, RAG citations, audited Memory, OKX market research, paper trading, strategy experiments, strategy knowledge memory, risk-gated live intents, OKX Testnet signed execution, and a BitPro MCP adapter for external data access plus non-live strategy lifecycle workflows.
Research output only. Nothing in this repository is investment advice.
cp .env.example .env
uv run pytest -q
npm exec --yes pnpm@10 -- -C frontend install
npm exec --yes pnpm@10 -- -C frontend dev
uv run uvicorn hypertrade.main:app --app-dir backend/src --host 0.0.0.0 --port 3334./scripts/check.shThis runs frontend install/lint/test/build, Python ruff, mypy, and pytest.
Start the standalone terminal Agent:
uv run hypertrade
# short alias
uv run htRun one prompt locally:
uv run hypertrade ask "请做行情归纳"Use the terminal harness against a running API:
uv run ht /login
uv run ht ask "请做行情归纳"/login saves the remote API URL, username, and password to
~/.hypertrade/client.env with local-only file permissions. After that,
uv run ht defaults to the saved remote API unless --local is passed.
Force local mode even when HYPERTRADE_API_URL is set:
uv run hypertrade --localIn interactive chat, slash commands inspect harness state without starting a new Agent run.
/help explains every command, and /tools shows each Agent tool with its category,
approval gate, and purpose:
Free-form Agent prompts show a live Thought / Thinking status block in interactive
terminals while the planner or tools are still running.
Interactive terminals also render Markdown reports into readable headings, lists, and tables;
set HYPERTRADE_RENDERER=plain when raw Markdown is needed for scripts.
Rich terminal output folds internal graph/preflight trace rows into a compact tool summary by
default; set HYPERTRADE_TRACE=full when a full audit trace is needed.
/help
/status
/tools
/runs
/memory
/strategy
/research 研究BTC趋势突破
/backtest
/backtest latest
/rag risk
/memory search risk
/model deepseek
/experiment 研究ETH趋势突破
/live execute loi_...
/evals
docs/README.md: documentation index and current capability map.docs/spec.md: product scope, acceptance criteria, and boundaries.docs/progress.md: latest implementation/deployment state.docs/architecture/: module-level architecture notes.docs/knowledge/tool-usage-guide.md: operator guide for Agent tools and validation.docs/runbooks/: deployment, BitPro MCP, smoke, backup, and incident procedures.