ai-market-maker

分类: 工具与效率 | 上传者: olaxbt | 下载: 0 | 版本: v1.0（最新）

在执行前使用**硬风险卫士否决**来操作和解释**多代理交易工作流程**。更喜欢**结构化跟踪**（“thought_process”、提案、否决）以实现透明度——与 Nexus 仪表板和“schema/agent_trace.json”保持一致。

更新日志: Source: GitHub https://github.com/olaxbt/ai-market-maker

目录结构

当前层级: tree/main/openclaw/

📄 manifest.json 1.5 KB
📄 SKILL.md 1.6 KB

SKILL.md

# Skill: AI Market Maker (Hedge Fund OS)

## Purpose

Operate and explain a **multi-agent trading workflow** with a **hard Risk Guard veto** before execution. Prefer **structured traces** (`thought_process`, proposals, veto) for transparency—aligned with the Nexus dashboard and `schema/agent_trace.json`.

## When to use

- Running or auditing the **LangGraph** pipeline (`src/main.py`).
- Mapping **personas** (`docs/personas/`) to **topology nodes** (`n1`–`n9`)—see `docs/persona_architecture_map.md`.
- Preparing **OpenClaw** tool calls against the future **NexusAdapter** (see `manifest.json` → `tools`).

## Boundaries

- **Secrets**: Never serialize untrusted dicts through LangChain `dumps`/`loads` with sensitive data; keep `langchain-core` patched (see `pyproject.toml`).
- **Live trading**: Execution is testnet-oriented today; production requires explicit mode flags and adapter work (roadmap P3/P4).
- **Veto**: Treat **Risk Guard** as **code-gated** (`risk_guard` node)—not a soft LLM suggestion.

## Key files

| Area | Location |
|------|----------|
| Workflow | `src/main.py` |
| State (target contract) | `src/schemas/state.py` |
| Flow events | `src/schemas/flow_events.py`, `src/flow_log.py` |
| Traces / UI contract | `schema/agent_trace.json`, `schema/nexus_payload.json` |
| Web dashboard | `web/` (Next.js) |

## Tool naming (future)

Use **stable names** from `openclaw/manifest.json` (`nexus.*`) so host runtimes can register MCP/OpenClaw tools without renames when the adapter lands.

## Versioning

Bump `manifest.json` → `version` when changing tool schemas or breaking payload fields.

登录后下载/点赞/收藏 ❤ 157 | ★ 0

ai-market-maker

目录结构

SKILL.md

举报内容

提示