Skip to content

README.en.md

Latest commit

 

History

History
82 lines (51 loc) · 3.33 KB

README.en.md

File metadata and controls

82 lines (51 loc) · 3.33 KB

Fubon NeoAPI Skill Refinement

Primary README (Chinese): README.md

This repository is used to test and refine the neoapi-python skill bundle for AI coding assistants (Codex, Claude Code). The skill provides guidance for the Fubon Neo Python SDK, including trading and market data workflows.

Repo Layout

  • skills/neoapi-python/ - The skill bundle (SKILL.md, references, llms*.txt, VERSION, INSTALL.md)
  • update-skill.ps1 - Windows update helper (pulls from GitHub and installs to ~/.codex/skills/public/)
  • update-skill.sh - macOS/Linux update helper

Compatibility

The skill is packaged as plain files with a SKILL.md entry point. This keeps the bundle portable across AI platforms that accept local skill bundles or custom instruction folders. For non-Codex platforms, set the install destination to that platform’s skill/instruction folder (use the INSTALL_DIR parameter or env var in the update scripts).

GitHub Repo

  • https://github.com/phenomenoner/neoapi-skill

Install

See skills/neoapi-python/INSTALL.md for full instructions. This repo uses skills/ (no leading dot). End users install to:

  • Windows: %USERPROFILE%\.codex\skills\public\neoapi-python
  • macOS/Linux: ~/.codex/skills/public/neoapi-python

Versioning

The skill version is stored in skills/neoapi-python/VERSION using semver. Current version: 1.0.0-beta.30 (Beta 1.0.0).

Multi-Agent Adapters

Root-level adapter files are provided for major agents, with Chinese-first guidance and English supplement:

  • CLAUDE.md
  • GEMINI.md
  • AGENTS.md

These adapters point to skills/neoapi-python/ as the canonical source.

LLM Docs (Online)

The official LLM-oriented pages and llms*.txt endpoints are available online and should be treated as primary references:

  • https://www.fbs.com.tw/TradeAPI/docs/welcome/build-with-llm
  • https://www.fbs.com.tw/TradeAPI/en/docs/welcome/build-with-llm/
  • https://www.fbs.com.tw/TradeAPI/llms.txt
  • https://www.fbs.com.tw/TradeAPI/llms-full.txt
  • https://www.fbs.com.tw/TradeAPI/en/llms.txt
  • https://www.fbs.com.tw/TradeAPI/en/llms-full.txt

The bundled llms*.txt files are offline snapshots. When the official endpoints change, refresh the repo copies and bump VERSION; otherwise the update scripts will treat the installed bundle as unchanged.

Updating from GitHub

The update scripts download the repo zip from GitHub, copy skills/neoapi-python into the install location, and compare VERSION to skip if already up to date.

Example:

.\update-skill.ps1 -Repo phenomenoner/neoapi-skill
./update-skill.sh phenomenoner/neoapi-skill

Local Regression Testing

  • Local integrated runner: .test/test_runner.py (not included in published skill bundle).
  • Logs are written to .test/logs/ as both text and JSON summaries.
  • Suites:
    • smoke: basic login + market/trade sanity checks
    • complex: multi-symbol marketdata matrix + dual-order lifecycle
    • all: full coverage (includes complex checks)

Changelog

The full release history is maintained in CHANGELOG.md.

The current latest version is 1.0.0-beta.30, with the main update being live fill-handling guidance: use set_on_filled as the low-latency primary fill path, get_order_results as periodic safe-net reconciliation, and never fall back to submitted limit price for P/L.