AIPriceAction

Financial data platform with AI-powered analysis for Vietnamese stocks, crypto, and global markets.

Tiếng Việt

---

Get started in 30 seconds

npx skills add quanhua92/aipriceaction

Then ask any AI agent:

"Compare VCB, TCB, MBB, CTG across 1h and 1D — which bank has the strongest trend?"

"Analyze VIC with volume profile and price action"

"Compare SJC gold price with global gold (GC=F)"

"Volume profile for BTCUSDT — where is the POC and value area?"

"Compare FPT, VNM, HPG with Wyckoff analysis"

"Show me top performers ranked by trading value"

"Deep research: which sectors are leading the VN market right now?"

Four skills are installed: aipa-data (raw OHLCV), aipa-analyze (AI-powered analysis), aipa-research (multi-agent deep research), and aipa-fundamentals (fundamental analysis with PE, ROE, NPL, CAR, screening). Works with Claude Code, Gemini CLI, and Codex.

No install — one file

Download AGENTS.md and drop it in your project root. Gemini CLI auto-detects it; for Claude Code, copy or symlink it:

# Option 1: Symlink (auto-updates when AGENTS.md changes)
ln -s AGENTS.md CLAUDE.md

# Option 2: Copy (manual update needed)
cp AGENTS.md CLAUDE.md

Any AI agent now has full aipa-cli documentation — raw data commands, analysis flags, caching rules, and when-to-use table — without installing skills. Requires Python on the machine — the AI auto-installs aipa-cli on first use. Web-only agents (e.g. Claude.ai web) won't work. Note: with copy, AGENTS.md must be downloaded again manually when updated, while skills update easily via npx skills update.

---

Install

I want to...	Install	One-liner
Use AI agent — no skill install	Download AGENTS.md	Auto-installs aipa-cli
Add AI agent skills	npx skills add quanhua92/aipriceaction	Auto-installs aipa-cli
Use the CLI / TUI	uv tool install aipa-cli	Terminal analysis
Build with Python	pip install aipriceaction	Pandas DataFrames

Updating

# Update AI agent skills
npx skills update

# Update CLI (if installed via uv tool)
uv tool upgrade aipa-cli

---

Featured capabilities

Volume Profile

POC, value area, and volume-by-price histogram from 1-minute data.

aipa volume-profile VCB
aipa volume-profile BTCUSDT --source crypto --bins 30
aipa volume-profile FPT --start-date 2026-05-05 --end-date 2026-05-09

Top Performers

Rank tickers by price change, volume, MA scores, money flow, or sector.

aipa performers --sort-by value --limit 5
aipa performers --sort-by ma50_score --group NGAN_HANG
aipa performers --sort-by total_money_changed --source crypto

AI Analysis

Wyckoff, VPA, and smart money signals with structured context.

aipa analyze VCB --interval 1D
aipa analyze VCB FPT VIC --interval 1h
aipa deep-research --run

---

Architecture

flowchart LR
    subgraph Sources ["📡 Data Sources"]
        VN["VCI · Vietstock · VNDirect<br/>VPS · DNSE<br/>VN Stocks"]
        BIN["Binance<br/>Crypto"]
        YAH["Yahoo Finance<br/>US / Global"]
        SJC["sjc.com.vn<br/>SJC Gold"]
    end

    subgraph Backend ["🦀 Rust Backend"]
        W["Background Workers"]
        PG[("PostgreSQL")]
        RD[("Redis")]
        API["REST API<br/>:3000"]
        W --> PG --> RD --> API
    end

    subgraph S3 ["🗄️ S3 Archive"]
        CSV["Daily CSV<br/>Yearly CSV<br/>+ tickers.json"]
    end

    subgraph Consumers ["👥 Consumers"]
        SDK["Python SDK<br/>pip install aipriceaction"]
        CLI["AIPA CLI<br/>uv tool install aipa-cli"]
        SKILLS["AI Agent Skills<br/>aipa-data · aipa-analyze · aipa-research · aipa-fundamentals"]
        WEB["aipriceaction.com<br/>Web Frontend"]
    end

    VN & BIN & YAH & SJC --> W
    API -- "sync every 60m" --> CSV
    CSV -- "plain HTTP<br/>no credentials" --> SDK
    API -- "live overlay" --> SDK
    SDK --> CLI --> SKILLS
    API --> WEB

Loading

---

Data sources

Market	Provider	Ticker examples	Intervals
Vietnamese stocks	VCI / Vietstock / VNDirect / VPS	VCB, FPT, VNINDEX	1m, 1h, 1D
US / intl. stocks	Yahoo Finance	AAPL, GOOGL, GC=F	1m, 1h, 1D
Cryptocurrency	Binance	BTCUSDT, ETHUSDT	1m, 1h, 1D
SJC gold	sjc.com.vn	SJC-GOLD	1D

Aggregated intervals (5m, 15m, 30m, 4h, 1W, 2W, 1M) are computed on-demand from base 1m/1D data.

---

Components

AI Agent Skills

Four skills for Claude Code, Gemini CLI, and Codex: aipa-data fetches raw OHLCV data, aipa-analyze runs AI-powered single/multi-ticker analysis with Wyckoff and VPA patterns, aipa-research runs a multi-agent supervisor/worker/reviewer pipeline for sector-wide deep research, and aipa-fundamentals provides fundamental analysis with financial ratios, company profiles, and multi-criteria screening. See skills/README.md.

AIPA Terminal

Textual-based TUI with streaming chat, thinking/reasoning display, autocomplete, and slash commands. Also ships CLI subcommands for non-interactive analysis, volume profile, performers, live data, and deep research. If you're not using uvx, ask your AI agent to update: "update aipa-cli to the latest version". See aipriceaction-terminal/README.md.

Python SDK

Reads OHLCV data from a public S3 archive via plain HTTP — no credentials needed. Returns pandas DataFrames with optional MA indicators and scores. Includes an AI Context Builder for LLM-powered analysis, LangChain/LangGraph agent integration, and live data overlay. See sdk/aipriceaction-python/README.md.

Rust Backend

Axum REST API with background workers that sync OHLCV data from multiple providers into PostgreSQL, served through a Redis edge cache. Deploys as a single Docker container. See aipriceaction/README.md.

Frontend

Human-facing web UI at aipriceaction.com. Source at aipriceaction-web.

---

Self-host

cd aipriceaction
cp .env.example .env
docker compose up -d

See aipriceaction/README.md for the full setup guide, build-from-source instructions, and production deployment with HAProxy.

---

Repository structure

aipriceaction/              Rust backend -- API server, background workers, PostgreSQL + Redis
sdk/
  aipriceaction-python/     Python SDK -- reads from S3 archive, no credentials needed
aipriceaction-terminal/     Python TUI and CLI for AI-powered ticker analysis
skills/                     Claude Code skills for market analysis workflows

---

Deep dives

Document	What it covers
DATA_FLOW.md	S3 archive, live API, cache freshness, and the merge pipeline
VOLUME_PROFILE.md	Volume-by-price algorithm: POC, value area, and why uniform distribution
PERFORMERS.md	Top/worst market rankings: metrics, MA scores, money flow
MULTI_AGENTS_ANALYSIS.md	AI agent architecture: single-agent analyze vs multi-agent deep-research pipeline

Supply chain security

Rust dependencies are managed with cargo-cooldown — a 14-day minimum publish age policy ensures freshly published crates are not picked up until they've been available long enough for maintainers and automated scanners to review. See aipriceaction/cooldown.toml for configuration.

cargo cooldown update   # instead of cargo update
cargo cooldown check    # instead of cargo check

Development

See CLAUDE.md for development guidelines, architecture details, and contributor instructions.

License

MIT