Introduction¶

Developers working on football-related content often encounter two types of trouble: First, fragmented data—fixtures from one site, xG from another, leading to inconsistent metrics. Second, asking a model to recall “what is Real Madrid’s average xG this season” often results in it confidently fabricating a number. QiuXiaoCe Football Data aims to solve this by turning authoritative, structured football match fact sources into a Skill, allowing agents to fetch data via APIs rather than answering from memory.

The following introduces what this Skill is, what it can do, how to install and invoke it, and some practical pitfalls you might encounter.

What is this¶

In one sentence: An official football big data and AI pre-match analysis skill package from QiuXiaoCe, designed for content creators, data analysts, and fans, providing structured match fact sources for LLM agents.

The producer and maintainer is the QiuXiaoCe team (www.qiuxiaoce.com). The Skill is published as a community source on SkillHub at the coordinate @user_e5e08891/qiuxiaoce-football-data, categorized as professional. Official URL:

https://api.skillhub.cn/user_e5e08891/qiuxiaoce-football-data

The package contains a standard SKILL.md main entry point plus several workflow branches: intent routing, data fetching scripts, local caching, content templates, backtesting templates, and self-update scripts. The main entry file is only responsible for understanding user intent and routing; it does not answer data questions directly from its own content.

It is worth noting a version discrepancy: The version listed on the SkillHub page is 2.8.0, while the latest version currently returned by the official version API is 2.16.0 (released on 2026-10-25). The official API should be treated as the source of truth.

Coverage and Core Capabilities¶

Data covers the Top 5 European leagues (Premier League, La Liga, Serie A, Bundesliga, Ligue 1), Champions League, Europa League, and over 80 major leagues globally. The official site publicly lists 88 match data items, including fixtures and scores, starting lineups, player and team statistics, advanced metrics, injuries/suspensions, and weather, with automatic daily statistics.

Capabilities are divided into two layers:

Free Layer (No Key Required): Real-time standings, team and league search (CN/EN), top scorers, and top assists.
Key Required Layer: Registering at the QiuXiaoCe Developer Console (https://www.qiuxiaoce.com/data-account/) grants 500 trial points, unlocking deeper data such as the Single Match Panorama Data Pack.

Metrics explicitly listed by the skill package include:

Team Advanced Profiles: xG, xGA, Tactical Fingerprint.
Player Advanced Ratings: Overall Rating, OTI Offensive Threat (OTI and OTI-XG metrics), DII Defensive Involvement, GIM Game Influence Model, plus Form, Stability, Fatigue, and xG (Expected Goals), xA (Expected Assists).
Single Match Shot Maps: Shot-by-shot coordinates, individual xG/PSxG, goalkeeper overperformance.
Others: Team Home Profile, Market Behavior Index, Injury/Suspension Lists, Head-to-Head History, Transfer Records, Manager Career History, Referee Style.

The Single Match Panorama Data Pack claims over 200 dimensions, covering attacking/defending averages, quantified loss from injuries, and H2H history. Skill trigger words include: check matches, today’s matches, tomorrow’s schedule, standings, top scorers, injuries, H2H, transfers, manager, referee, pre-match analysis, help me predict, generate report, check quota, update Skill.

Installation and Activation¶

The official installation does not provide a one-line command. Instead, it follows a process of “Fetch Version Info → Download → Verify → Extract → Configure Key,” all provided within code blocks.

Step 1: Fetch the latest version info, returning version, download_url, sha256 checksums, file list, and release notes:

curl -s "https://www.qiuxiaoce.com/wp-json/abv2-creator/v1/skill/version"

Step 2: Download the skill package using the returned download_url and verify each file against the returned sha256. The download address for the current version looks like:

curl -LO "https://www.qiuxiaoce.com/wp-json/abv2-creator/v1/skill/download?v=2.16.0"

Step 3: Extract to the local skill directory, preserving the package’s top-level directory structure, and complete initialization according to the SKILL.md inside the package. Package files include SKILL.md, scripts/ (fetch_match.py, query_backtest.py, check_quota.py, update_skill.py, local_store.py), prompts/, references/, templates/, and workflows/.

Step 4: Configure the API Key. The key format is qxck_xxxxxxxx. It can be written to the environment variable QIUXIAOCE_API_KEY or written to a local .qiuxiaoce_key file by the skill. Free layer APIs do not require a key, so you can test those first before configuring.

Step 5: Environment Self-Check. The scripts depend on Python (≥3.8, using only the standard library, no pip install needed). Confirm the environment before running scripts for the first time in each session:

python3 --version || python --version

If both commands fail, do not attempt to fix the installation yourself; directly inform the host AI tool that a Python 3 environment is required and needs to be configured in the PATH. Initialize the local database (optional, it will be created automatically on the first call):

python3 scripts/local_store.py init

Environmental Differences: The official statement clarifies that installation, configuration, and execution environments for third-party clients like WorkBuddy, Qoder, and Trae, as well as how to import and load Skills within clients, fall under the scope of the third-party software’s own general skills. QiuXiaoCe does not provide free manual consultation or technical support for these. Environments that can only chat via web dialogs and cannot access the file system (such as Doubao, Tongyi Qianwen, Kimi, etc.) cannot install the Skill. The official alternative is to provide the AI with your API Key directly and perform data analysis according to the API documentation.

Typical Usage¶

API Root URL:

https://www.qiuxiaoce.com/wp-json/abv2-creator/v1

Authentication is via the X-API-Key header. Points are deducted per request, but duplicate requests within 60 seconds are not charged again. The minimal reproducible example provided in the official docs is the Today’s Overview:

curl -H "X-API-Key: YOUR_KEY_HERE" \
  "https://www.qiuxiaoce.com/wp-json/abv2-creator/v1/today/digest"

If you want to read the documentation before calling the API, use the public documentation interface (no key required):

# Get full documentation index, returns titles and slugs for all chapters
curl -s "https://www.qiuxiaoce.com/wp-json/abv2-creator/v1/docs"

# Read single chapter content
curl -s "https://www.qiuxiaoce.com/wp-json/abv2-creator/v1/docs/{slug}"

Human-readable pages are at https://www.qiuxiaoce.com/creator-docs/{slug}/. The complete API documentation is at https://www.qiuxiaoce.com/data-docs/ (48 data interfaces, each labeled with purpose, parameters, price per call, and real response examples).

The data fetching logic is “Local First”: check the local SQLite first, and only call the API if there is a miss or the data is expired. Important data obtained from the API must be persisted to the database. If you encounter errors, first check the official troubleshooting manual https://www.qiuxiaoce.com/creator-docs/api-error-codes-troubleshooting/. Common codes:

401: Key missing, incorrect format, or has been reset/revoked (api_key_revoked). Go to the console to copy the latest complete Key and sync updates to all scripts and clients configured with this Key.
402: Insufficient points or expired. Recharge at https://www.qiuxiaoce.com/data-service/.
403: Interface not open, or insufficient permissions (some match reports require subscription).
404: Entity ID or interface path written incorrectly.
429: Request too fast. Wait according to the Retry-After response header and retry.

Check remaining points before making large batch calls to avoid failure halfway. The skill package also has a built-in update process (update_skill.py). Before upgrading, it compares the official new version with your modified files item by item, allowing you to decide whether to use the new version directly or keep changes and merge the upgrade.

Use Cases and Notes¶

Suitable for three types of people: Content creators in football media (pre-match reports, copy generation), Data analysts doing scientific modeling (requiring structured data with verifiable metrics), and Developers who want to connect match fact sources to agents.

Several limitations and pitfalls:

Hard Environment Prerequisites: Requires a desktop or terminal-based AI tool capable of reading/writing local files and executing commands, and requires Python 3. Pure web dialog interfaces cannot use the Skill itself.
Point-based Billing: Points are deducted per call, with no duplicate deduction for the same request within 60 seconds. The free layer only includes standings, team/league search, top scorers, and top assists. Deeper data like the Single Match Panorama Pack consumes points. Registration grants 500 trial points.
Calibrate Metrics First: The skill requires calibrating metrics like xG, xGA, OTI, and DII according to official documentation before use. Insufficient sample sizes must be noted in the conclusion. If data is missing, state what is missing; do not gloss over it with “approximately” or “maybe.”
Version Discrepancy: The version shown on the SkillHub list page (2.8.0) lags behind the latest release on the official API (2.16.0). Verify the actual version number after installation.
Service Boundary: The official statement explicitly declares no betting advice is provided, nor are accuracy guarantees made. The content of the skill package is for match research, data science modeling, and football fan communication only, and does not constitute any betting or investment advice.
Data Degradation Awareness: Recent versions have added degradation package labels: The Single Match Panorama Pack now includes meta.quality (complete/degraded). Degraded packages automatically refund points. Re-fetching after the backend rebuilds will yield the complete version.

Conclusion¶

The value of QiuXiaoCe Football Data lies in packaging “Authoritative + Structured + Verifiable Metrics” into a Skill that agents can call directly, saving the effort of connecting multiple data sources and reconciling metrics yourself. You can try the free layer first, and configure a Key when you need deeper data.

Official Website: https://www.qiuxiaoce.com
SkillHub Page: https://www.skillhub.cn/skills/user_e5e08891/qiuxiaoce-football-data
API Documentation: https://www.qiuxiaoce.com/data-docs/

QiuXiaoCe Football Data: Connecting AI Agents to Authoritative Football Data Sources