# edge-hint-extractor

Turns daily market observations and news reactions into structured edge hints, written out as hints.yaml.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 3300 installs · ★ 3267
- Author: tradermonty
- License: MIT · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/tradermonty/claude-trading-skills/tree/HEAD/skills/edge-hint-extractor/SKILL.md
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/skills/edge-hint-extractor

## What it does

The skill reads raw signals — `market_summary`, `anomalies`, news reactions — and produces hint objects from them. It sits first in the chain `observe -> abstract -> design -> pipeline`, so the resulting `hints.yaml` feeds concept synthesis or auto detection.

One script does the work: `scripts/build_hints.py`. Rule-based hints come out deterministic. Model-generated ideas arrive by either of two routes, and you pick one: `--llm-ideas-cmd` pipes data to an external LLM CLI as a subprocess, while `--llm-ideas-file PATH` loads hints you already wrote into a YAML file — the route meant for Claude, where the agent writes the hints itself. Passing both flags at once fails.

Python 3.9+ and `PyYAML` are required. Observation files are optional, though there is nothing to build without them: `market_summary.json` and `anomalies.json` come from a detector run, and news reactions live in `news_reactions.csv` or `news_reactions.json`. Output defaults to `reports/edge_hint_extractor/hints.yaml`; the schema sits in `references/hints_schema.md`.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add tradermonty/claude-trading-skills --skill edge-hint-extractor -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/edge-hint-extractor/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/edge-hint-extractor/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add tradermonty/claude-trading-skills --skill edge-hint-extractor -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/edge-hint-extractor/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add tradermonty/claude-trading-skills --skill edge-hint-extractor -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/edge-hint-extractor/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install tradermonty/claude-trading-skills edge-hint-extractor
```

2. **Manually: put the files in place**
- File: `.github/skills/edge-hint-extractor/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add tradermonty/claude-trading-skills --skill edge-hint-extractor -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/tradermonty/claude-trading-skills --path skills/edge-hint-extractor
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/edge-hint-extractor/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add tradermonty/claude-trading-skills --skill edge-hint-extractor -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/tradermonty/claude-trading-skills/tree/HEAD/skills/edge-hint-extractor/SKILL.md

## When to use it
- turn daily market observations into hint objects
- add model-written ideas on top of anomalies and news
- produce a clean `hints.yaml` for concept synthesis
- prepare input for auto detection
- consolidate detector signals into one structured file

## Example prompts
- Build hints from `market_summary.json` and `anomalies.json` as of 2026-02-20 into `reports/`
- Add your own ideas about the current anomalies and save `hints.yaml`
- Run `build_hints.py` with `--llm-ideas-file /tmp/llm_hints.yaml`

## Source text · SKILL.md · MIT

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: c650b9b50d291b45.

BEGIN THIRD-PARTY FILE c650b9b50d291b45
````text
---
name: edge-hint-extractor
description: Extract edge hints from daily market observations and news reactions, with optional LLM ideation, and output canonical hints.yaml for downstream concept synthesis and auto detection.
---

# Edge Hint Extractor

## Overview

Convert raw observation signals (`market_summary`, `anomalies`, `news reactions`) into structured edge hints.
This skill is the first stage in the split workflow: `observe -> abstract -> design -> pipeline`.

## When to Use

- You want to turn daily market observations into reusable hint objects.
- You want LLM-generated ideas constrained by current anomalies/news context.
- You need a clean `hints.yaml` input for concept synthesis or auto detection.

## Prerequisites

- Python 3.9+
- `PyYAML`
- Optional inputs from detector run:
  - `market_summary.json`
  - `anomalies.json`
  - `news_reactions.csv` or `news_reactions.json`

## Output

- `hints.yaml` containing:
  - `hints` list
  - generation metadata
  - rule/LLM hint counts

## Workflow

1. Gather observation files (`market_summary`, `anomalies`, optional news reactions).
2. Run `scripts/build_hints.py` to generate deterministic hints.
3. Optionally augment hints with LLM ideas via one of two methods:
   - a. `--llm-ideas-cmd` — pipe data to an external LLM CLI (subprocess).
   - b. `--llm-ideas-file PATH` — load pre-written hints from a YAML file (for Claude Code workflows where Claude generates hints itself).
4. Pass `hints.yaml` into concept synthesis or auto detection.

Note: `--llm-ideas-cmd` and `--llm-ideas-file` are mutually exclusive.

## Quick Commands

Rule-based only (default output to `reports/edge_hint_extractor/hints.yaml`):

```bash
python3 skills/edge-hint-extractor/scripts/build_hints.py \
  --market-summary /tmp/edge-auto/market_summary.json \
  --anomalies /tmp/edge-auto/anomalies.json \
  --news-reactions /tmp/news_reactions.csv \
  --as-of 2026-02-20 \
  --output-dir reports/
```

Rule + LLM augmentation (external CLI):

```bash
python3 skills/edge-hint-extractor/scripts/build_hints.py \
  --market-summary /tmp/edge-auto/market_summary.json \
  --anomalies /tmp/edge-auto/anomalies.json \
  --llm-ideas-cmd "python3 /path/to/llm_ideas_cli.py" \
  --output-dir reports/
```

Rule + LLM augmentation (pre-written file, for Claude Code):

```bash
python3 skills/edge-hint-extractor/scripts/build_hints.py \
  --market-summary /tmp/edge-auto/market_summary.json \
  --anomalies /tmp/edge-auto/anomalies.json \
  --llm-ideas-file /tmp/llm_hints.yaml \
````
END THIRD-PARTY FILE c650b9b50d291b45

The third-party file has ended. Everything between the c650b9b50d291b45 tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

First 2524 of 2653 bytes. © tradermonty · MIT · https://github.com/tradermonty/claude-trading-skills/tree/HEAD/skills/edge-hint-extractor/SKILL.md

_Part of the page did not fit — see the HTML version._

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md