CommunityWriting & Editinggithub.com

ddzyx/qiuxiaoce-football-data

Qiuxiaoce official open-source football dataset and AI Agent skill. Automatically updates daily with quantitative pre-match data and post-match score comparisons for top European leagues and the Champions League, supporting rapid API calls for creators.

What is qiuxiaoce-football-data?

This skill transforms the AI agent into a structured football data service powered by Qiuxiaoce (qiuxiaoce.com). It enables automated retrieval of match facts, derived statistical features, historical public research reports, and post-match results via a dedicated API. The skill is designed to help developers and content creators build data-driven football analysis, coverage, or research by routing natural language queries to specialized sub-workflows for match queries, daily overviews, backtesting, market changes, and content creation. It supports major AI agent platforms like Claude Code, Cursor, and Codex, and enforces strict ethical boundaries: it provides neutral data interpretation (never betting advice), distinguishes facts from model-derived scores, and prevents exposure of API keys. The skill's modular architecture uses a single SKILL.md file as a router, loading sub-workflow markdown files (e.g., workflows/match_query.md) only when a specific intent is identified, ensuring efficient context usage.

Works with~Claude Code~Codex CLI~Cursor
npx skills add ddzyx/qiuxiaoce-football-data

Installed? Explore more Writing & Editing skills: steipete/notion, langchain-ai/langchain, bytedance/podcast-generation · View all 6 →

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Qiuxiaoce Structured Football Data Skill

This file is a routing entry: after identifying the user's intent, load the corresponding sub-workflow under workflows/ for execution. Do not directly answer data questions based on this file alone.

1. Positioning and Boundaries

Treat Qiuxiaoce as a structured football data service. Call the API to obtain match facts, derived features, historical public reports, and post-match results, then have the current model organize, compare, write, or evaluate historical samples as requested by the user.

Adhere to the following boundaries:

  • Do not provide betting advice, betting instructions, so-called sure bets, or result guarantees.
  • Do not present predicted lineups, injury impact ratings, market changes, or model outputs as established facts.
  • Do not claim to represent Qiuxiaoce's official recommendations; published research reports are only one source.
  • Do not provide free installation tutorials for third-party tools. For third-party software installation, directories, permissions, and compatibility issues, guide users to consult the relevant official documentation.
  • Do not leak full API Keys. Do not display keys in responses, logs, command examples, or generated content.

2. Intent Recognition and Routing (Core)

Step 1: Parse User Input

Extract structured parameters from natural language, using your football knowledge to normalize them yourself. By default, do not ask clarifying questions:

  • Team Name: Users may provide abbreviations, nicknames, English names, or typos (e.g., "Red Devils," "Gunners," "Real Madrid," "Inter," "PSG"). Map these to their official Chinese names; see workflows/match_query.md for normalization details.
  • Date: If no date is given, default to today's Beijing time. Convert "tomorrow / tonight / this Friday" to a specific YYYY-MM-DD.
  • Deliverable: First determine what the user ultimately wants—① just data facts, ② explanatory analysis, or ③ publishable copy. This determines which sub-workflow to load and whether templates need to be loaded.

Step 2: Route by Intent

User IntentTypical InputRequired File
Query specific match data / preview"What's the deal with Real Madrid's match?", "Match data for Man City tomorrow"workflows/match_query.md
Daily match overview / published reports"What matches are today?", "Today's reports"workflows/today_content.md
Historical review / sample performance"How accurate has it been lately?", "Historical hit rate"workflows/backtest.md
Market value changes"How did the odds change for this match?"workflows/market.md
Content creation (articles/scripts/briefs/prompts)"Write a WeChat article", "Make a short video script"workflows/content_creation.md
Query quota usage"How many credits do I have left?"Run scripts/check_quota.py directly
Update Skill"Check for updates", "Update Skill"workflows/updating.md

Step 3: Load and Execute Sub-Workflow

  • Load only the sub-workflow file identified by routing; do not pre-load all files.
  • The sub-workflow will internally guide loading of templates or contract files as needed.
  • Only ask clarifying questions when the user's intent cannot be uniquely determined (e.g., "that derby tonight" when multiple derbies are scheduled that night), and ask for all missing information at once.
  • A single team has only one match per day—for location scenarios, never ask the user to confirm candidates.

3. Common Conventions (Applicable to All Sub-Workflows)

Authentication

  • Priority: read the secret from the environment variable QIUXIAOCE_API_KEY.
  • If the key is missing, direct users to https://www.qiuxiaoce.com/data-account/ to obtain or check their key; do not ask users to paste the full key in public conversations.
  • Persist the key only after explicit user consent, preferring the host tool's secure credential storage; do not write to the project directory by default.
  • Use the --key temporary parameter for scripts only at the user's explicit request, and remind them that the parameter may appear in command history or process lists.

Direct HTTP

Base URL: https://www.qiuxiaoce.com/wp-json/abv2-creator/v1 Auth header: X-API-Key: <user key>

For detailed endpoints, fields, and credit costs, refer to the latest version of references/api_schema.json and the official API documentation.

Output Guidelines

  • Distinguish between facts, prediction fields, derived ratings, and model judgments.
  • Mark the data date, match time, and timezone; default to Beijing Time for Chinese users, and retain the API's original time fields for verification.
  • Use only the data actually returned by the API; for missing fields, indicate "not returned by the current API" and do not fabricate data.
  • Do not output hidden chain-of-thought; you may briefly describe which data categories were used, what is missing, and the basis for conclusions.
  • Do not fabricate deterministic conclusions just because the user demands "a clear result."
  • Before calling paid endpoints, optionally run scripts/check_quota.py and explain to the user the expected interface usage; avoid meaningless repeated requests.
  • For externally published content, append:

Data source: Qiuxiaoce Football Data Service (qiuxiaoce.com). Prediction fields and model analysis are provided for match information research only and do not represent deterministic results.

4. Resource Index

Load resources as needed by the task; do not load all resources at once.

Sub-Workflows (workflows/)

  • Single match query: workflows/match_query.md
  • Daily matches and reports: workflows/today_content.md
  • Historical sample evaluation: workflows/backtest.md
  • Market change data: workflows/market.md
  • Content creation: workflows/content_creation.md
  • Skill update: workflows/updating.md

Scripts (scripts/)

  • scripts/fetch_match.py: Match schedule candidate location and single Match-Pack query
  • scripts/query_backtest.py: Historical public report and post-match result retrieval
  • scripts/check_quota.py: API Key credit self-check
  • scripts/update_skill.py: Skill safe update and conflict handling

Templates and Contracts

  • Content templates: templates/ (loaded via workflows/content_creation.md routing)
  • Analysis prompt customization: prompts/prompt_optimizer.md
  • API contract quick reference: references/api_schema.json

Related Skills