功能描述

Multi-source flight aggregation — tickets, nonstop, round-trip, cabin. 航班机票/零登录/零API, zero login, zero account, zero API key. Pure Python, no browser. 机票价格/航...

使用说明 (SKILL.md)

FlyClaw ✈️ - 航班信息聚合查询工具 / Flight Information Aggregation Tool

Name: Flyclaw Bak
Author: qizha

概述 / Overview

五源航班聚合查询 — 零登录、零账号、零 API Key，轻量 Python 实现。 5-source flight aggregation — zero login, zero account, zero API key. Lightweight Python, no browser automation.

基于多源聚合架构，通过开源库及免费公开 API 获取航班动态、价格、时刻、实时位置等信息。支持中英文城市名和 IATA 代码输入。

Multi-source aggregation via open-source libraries and free public APIs for flight dynamics, prices, schedules, and real-time positions. Supports Chinese/English city names and IATA codes.

GitHub：https://github.com/AI4MSE/FlyClaw

零 API Key 依赖 / Zero API Key

无需注册任何账号或提供任何 API Key 即可使用全部核心功能。用户本地化掌控所有，程序不收集、不存储任何用户个人信息。同时规避浏览器模拟等复杂、不可靠和低效问题。

No registration or API key required. All data stays local — no personal data collected or stored. No browser automation overhead.

触发方式 / Trigger

用户说"查航班 CA981"、"上海飞纽约多少钱"、"PVG 到 JFK 明天的航班"、"往返机票上海到新加坡"、"商务舱北京到伦敦"、"所有航班包括转机"、"直飞" 等即可自动执行。默认查询行为是直飞+经济舱。

Trigger when user says "query flight CA981", "flights from Shanghai to New York", "round-trip PVG to SIN", "business class Beijing to London", "nonstop flights", "all flights including connecting", etc. Default behavior is nonstop + economy.

智能转换规则 / Smart Conversion Rules：

用户说"所有航班"/"包括转机"/"包括非直飞" → --stops any
用户说"直飞"/"不要转机" → --stops 0（默认）
用户说"最多一次中转" → --stops 1
用户说"最多两次中转" → --stops 2

数据来源 / Data Sources

飞猪 (Fliggy)：国内外航班价格（中国国内覆盖最全，人民币计价）
Google Flights：国内外国际航班价格、时刻
Skiplagged：国内外国际航班价格、时刻
FlightRadar24：航班动态、实时状态、延误、机型
Airplanes.live / ADSB.lol：ADS-B 实时位置

多源并发查询，智能合并互补。插件式架构，支持无限扩展——每个数据源为独立模块。特别感谢以上公开数据源为公益和大众需求提供的便利！

功能 / Features

按航班号查询：查询指定航班的动态信息（状态、时刻、延误、机型等）
按航线搜索：查询两地之间的航班列表（含价格、经停、时长等）
城市级搜索：输入城市名自动搜索该城市所有机场（如"上海"→PVG+SHA）
高级搜索：往返搜索、多旅客配置、舱位选择、排序、直飞过滤
中英文输入：支持中文城市名、英文名、IATA 代码，7912 机场全覆盖

重要：输出格式与多日查询 / Output Format & Multi-Day Queries

默认输出为 JSON（stdout），直接 json.loads() 即可解析，示例：

[{"flight_number": "CA981", "price": 472.0, "origin_iata": "PVG", "destination_iata": "GVA", ...}]

无结果时返回 []。错误和日志仅输出到 stderr，不影响 JSON 解析。价格默认为人民币（CNY），每条记录含 currency 字段标注货币。可用 --currency usd 统一转换为美元，或 --currency cny（默认）。汇率可在 config.yaml 中配置（默认 7.25）。可用 -o table 切换为人类可读表格。

多日查询：search 命令每次只查一天。查询一周最低价等场景，需拆成多个日期并发执行，分别获取 JSON 结果后自行合并比较。

调用方式 / Usage

按航班号查询

python flyclaw.py query --flight CA981

按航线搜索

python flyclaw.py search --from 上海 --to 纽约 --date 2026-04-01

往返搜索

python flyclaw.py search --from PVG --to LAX --date 2026-04-15 --return 2026-04-25

商务舱 + 2 人

python flyclaw.py search --from PVG --to JFK --date 2026-04-15 --cabin business -a 2

直飞 + 按价格排序

python flyclaw.py search --from PVG --to SIN --date 2026-04-15 --stops 0 --sort cheapest

包含经停航班

python flyclaw.py search --from PVG --to JFK --date 2026-04-15 --stops any

按日期过滤查询结果

python flyclaw.py query --flight CA981 --date 2026-04-01
python flyclaw.py query --flight CA981 --date today

关闭智能查价

默认启用智能查价，会自动在航班号查询时补充价格信息。关闭后可节约查询时间。

python flyclaw.py query --flight CA981 --no-relay

搜索参数

参数	短标志	默认值	说明
`--from`	—	（必填）	出发地
`--to`	—	（必填）	目的地
`--date` / `-d`	—	—	出行日期 YYYY-MM-DD
`--return` / `-r`	—	—	返程日期（启用往返搜索）
`--adults` / `-a`	—	1	成人旅客数
`--children`	—	0	儿童旅客数
`--infants`	—	0	婴儿旅客数
`--cabin` / `-C`	—	economy	economy/premium/business/first
`--limit` / `-l`	—	不限制	最大结果数（不指定则返回全部）
`--sort` / `-s`	—	—	cheapest/fastest/departure/arrival
`--stops`	—	0	经停：0=直飞/1/2/any=不限

通用参数

-o table：表格格式输出（默认为 JSON）
-v：详细模式，显示数据来源和舱位

输入示例 / Input Examples

用户说	解析为	说明
"上海"	PVG + SHA	城市级：搜索所有上海机场
"PVG"	PVG	精确到浦东机场
"浦东"	PVG	别名精确匹配
"纽约"	JFK + EWR + LGA	城市级：搜索所有纽约机场
"北京"	PEK + PKX + NAY	城市级：搜索所有北京机场
"Shanghai"	PVG + SHA	英文城市名同样支持

安装配置 / Installation

pip install requests pyyaml curl_cffi flights
# 注意：不要安装 mcp、fast-flights、playwright 等调试模块，会增加安装时间且普通使用不需要

文件位置：主程序 flyclaw.py，配置 config.yaml，机场缓存 cache/airports.json。

依赖：Python 3.11+、requests（Apache-2.0）、pyyaml（MIT）、curl_cffi（MIT）、flights（MIT）。

安全性 / Security

零 API Key 依赖：程序运行不需要用户提供任何 API Key 或注册任何账号
程序不收集、不存储任何用户个人信息
所有网络请求仅用于查询公开航班数据

免责声明 / Disclaimer

本工具基于多源聚合架构，通过开源库及免费公开 API 获取数据
仅供学习研究用途，请遵守当地法律法规
Google Flights 在部分地区可能不可用
价格数据来自多个数据源，不同来源的价格可能存在差异（含税/不含税、舱位差异等），仅供参考

作者 / Author

公益技能，免费开源。 / Community-driven, open-source, free for everyone.

Email: [email protected]
小红书 / Xiaohongshu: @深度连接
GitHub: AI4MSE

许可证 / License：Apache-2.0

安全使用建议

This skill appears to implement a legitimate multi-source flight aggregator, but there are two things to check before installing: 1) Embedded/default credentials: config.yaml and README include comments such as 'api_key: "" # Leave blank uses built-in default key' and 'sign_secret: "" # Leave blank uses built-in default secret'. That implies the package may contain baked-in credentials or rely on a third party's account. Inspect the source files (especially any 'sources/*' modules) for hardcoded API keys, secrets, or opaque relay endpoints. If keys are present, ask the author who owns them and whether queries will be billed or rate-limited through a remote account. 2) Origin and install: the package contains many source files and large cache files but the registry entry has no install spec. Prefer installing from the upstream GitHub repo (https://github.com/AI4MSE/FlyClaw) or reviewing the exact files you will run. Run it in an isolated environment (container or dedicated VM) the first time, and monitor outgoing network connections to understand which external services the skill contacts (Fliggy, Google Flights/fli, Skiplagged, FR24, ADSB providers). Other practical advice: if you are uncomfortable with unknown built-in keys, set any api_key/sign_secret/serpapi_key fields to empty and disable features that require them (or replace with your own credentials). Consider reducing surface area by disabling 'route_relay' or unneeded sources in config.yaml. If you need higher assurance, ask the maintainer for a signed release or scan the source for hardcoded secrets and remote endpoints.

功能分析

Type: OpenClaw Skill Name: flyclaw-bak Version: 1.0.0 FlyClaw is a flight information aggregator that queries multiple public sources (Fliggy, Google Flights, FlightRadar24, etc.) to provide flight status and pricing. The code is well-structured, follows its stated purpose, and lacks evidence of malicious intent such as data exfiltration or unauthorized persistence. While it includes a feature in `airport_manager.py` to update airport data from a remote URL, it implements a validation check (`_validate_airport_data`) to ensure the downloaded content matches the expected schema, mitigating risks of arbitrary data injection. Hardcoded credentials in `sources/fliggy_mcp.py` are documented as public defaults from an upstream open-source project.

能力标签

cryptocan-make-purchases

能力评估

⚠ Purpose & Capability

Name/description claim 'zero API key / zero login' and 'no account required' but config and README contain fields for serpapi_key, fliggy_mcp api_key and sign_secret with comments like 'leave blank uses built-in default key' — this suggests the code may include or use default credentials for upstream services. That capability (including embedded/unknown credentials) is not justified by the 'zero API key' marketing claim and is disproportionate without explanation.

ℹ Instruction Scope

SKILL.md and READMEs are focused on flight queries, JSON output, and usage examples — instructions do not ask the agent to read unrelated system files or secrets. The tool supports updating airports from arbitrary URLs (update-airports --url), which is a normal feature but can download and overwrite local caches if misused. No explicit instructions to access environment variables are present.

ℹ Install Mechanism

The registry metadata lists 'No install spec — this is an instruction-only skill' but the package contains many Python source files (including large caches). That mismatch is an incoherence: code will have to be placed/run on the host but there's no declared install step. Dependencies are standard Python libs (requests, pyyaml, curl_cffi, flights) — no remote binary downloads detected in the metadata.

⚠ Credentials

The skill declares no required environment variables, which is good, but config.yaml exposes optional fields for third‑party credentials (serpapi_key, fliggy_mcp api_key & sign_secret) and explicitly says 'leave blank uses built-in default key' — meaning credentials or secrets may be embedded in code or used by default. That raises proportionality and provenance questions: why include default keys, who owns them, and could queries be routed through an account outside the user's control?

✓ Persistence & Privilege

No special persistence privileges requested (always: false). The skill does maintain local caches (cache/airports.json) and supports updating them; it doesn't request to modify other skills or system-wide agent settings according to the provided metadata.

版本历史

v1.0.0

Initial public release of flyclaw-bak: - Multi-source flight search and aggregation (5 major sources included). - No login, account, or API key required; all core features work out-of-the-box. - Supports search by flight number, route, city (Chinese/English/IATA), round-trip, cabin class, and direct/connecting flights. - Outputs results in JSON by default; table format available. - No personal data collected or stored; open source under Apache-2.0. - Fully documented CLI usage, parameters, and installation instructions.

元数据

Slug flyclaw-bak

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题

Flyclaw Bak 是什么？

Multi-source flight aggregation — tickets, nonstop, round-trip, cabin. 航班机票/零登录/零API, zero login, zero account, zero API key. Pure Python, no browser. 机票价格/航... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 122 次。

如何安装 Flyclaw Bak？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install flyclaw-bak」即可一键安装，无需额外配置。

Flyclaw Bak 是免费的吗？

是的，Flyclaw Bak 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

Flyclaw Bak 支持哪些平台？

Flyclaw Bak 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Flyclaw Bak？

由 qizha（@qizha）开发并维护，当前版本 v1.0.0。

Flyclaw Bak