/cfb-data

name: cfb-data
description: |
  College Football (CFB) data via ESPN public endpoints and the NCAA's official endpoints — scores, standings, rosters, schedules, game summaries, play-by-play, rankings, injuries, futures, team/player stats, and news for FBS, plus official FCS scoreboards, NCAA game detail with drive context, and the schools index. Zero config, no API keys.

  Use when: user asks about college football scores, standings, rankings, team rosters, schedules, game results, play-by-play, injuries, betting futures, team/player statistics, CFB news, or FCS games.
  Don't use when: user asks about NFL (use nfl-data), college basketball (use cbb-data), or non-sports topics.
license: MIT
# `metadata` is the only free-form field the VS Code agent-skills
# schema accepts; we nest the Machina manifest under `metadata.machina`
# so the picker (Machina Factory `/c`) and Truth Point can read it
# without breaking any other consumer of this SKILL.md. All
# `metadata.machina.*` fields are optional — older skills without
# them still parse and just render fewer chips in the picker.
metadata:
  author: machina-sports
  version: "0.1.0"
  machina:
    categories: [sports-data, college-football]
    pricing_tier: free
    # Connector slugs this skill needs at runtime. Empty = uses only
    # public/free APIs (ESPN here).
    integrations: []
    # Credential vault keys the customer must have set before invoking.
    # Empty = no auth required.
    vault_keys: []
    # Surface area for the picker + the agent's pre-flight. Keep
    # summaries under ~80 chars so the picker's expand panel stays
    # compact.
    commands:
      - name: get_scoreboard
        summary: Live or recent CFB scores. Filter by date, week, conference, or limit.
        returns: events[]
      - name: get_standings
        summary: Standings by conference. Use the `group` param for SEC, ACC, etc.
        returns: conferences[]
      - name: get_teams
        summary: All 750+ FBS teams with id, name, abbreviation, logo, location.
        returns: teams[]
      - name: get_team_roster
        summary: Full roster for a team — players, positions, jerseys, height/weight.
        returns: athletes[]
      - name: get_team_schedule
        summary: Season schedule for one team — opponent, date, score (if played), venue.
        returns: events[]
      - name: get_game_summary
        summary: Detailed box score, scoring plays, and leaders for a single game.
        returns: "{ game_info, competitors, boxscore, scoring_plays, leaders }"
      - name: get_rankings
        summary: AP Top 25, Coaches Poll, and CFP rankings — rank, previous, record, votes.
        returns: polls[]
      - name: get_news
        summary: CFB news articles, optionally filtered by team.
        returns: articles[]
      - name: get_play_by_play
        summary: Full drive + play-by-play breakdown for a game.
        returns: drives[]
      - name: get_schedule
        summary: Season schedule by week — filter by conference group.
        returns: events[]
      - name: get_injuries
        summary: Injury reports across every team.
        returns: teams[]
      - name: get_futures
        summary: Futures markets (National Championship, Heisman, etc.).
        returns: futures[]
      - name: get_team_stats
        summary: Team statistical profile by category — value, rank, per-game.
        returns: stats[]
      - name: get_player_stats
        summary: Player statistical profile.
        returns: stats[]
    # The agent in the customer build sandbox can call `machina sports
    # cfb <command> --json` directly to capture a live sample output —
    # we don't ship static fixtures here so the public repo stays
    # data-free. List the runtime invocation pattern explicitly so
    # tooling / pre-flight knows how to fetch a real shape on demand.
    runtime:
      cli: "machina sports cfb"
      sample_command: "machina sports cfb get_scoreboard"
    references:
      api: references/api-reference.md
      conferences: references/conference-ids.md
      teams: references/team-ids.md

College Football Data (CFB)

Before writing queries, consult `references/api-reference.md` for endpoints, conference IDs, team IDs, and data shapes.

Setup

Before first use, check if the CLI is available:

which sports-skills || pip install sports-skills

If `pip install` fails with a Python version error, the package requires Python 3.10+. Find a compatible Python:

python3 --version  # check version
# If < 3.10, try: python3.12 -m pip install sports-skills
# On macOS with Homebrew: /opt/homebrew/bin/python3.12 -m pip install sports-skills

No API keys required.

Quick Start

Prefer the CLI — it avoids Python import path issues:

sports-skills cfb get_scoreboard
sports-skills cfb get_rankings
sports-skills cfb get_standings --group=8

CRITICAL: Before Any Query

CRITICAL: Before calling any data endpoint, verify:

• Season year is derived from the system prompt's `currentDate` — never hardcoded.
• For standings, the `group` parameter is set to the correct conference ID (see `references/api-reference.md`).
• If only a team name is provided, use `get_teams` to resolve the team ID.

Choosing the Season

Derive the current year from the system prompt's date (e.g., `currentDate: 2026-02-28` → current year is 2026).

• **If the user specifies a season**, use it as-is.
• **If the user says "current", "this season", or doesn't specify**: The CFB season runs August–January. If the current month is February–July (offseason), use `season = current_year - 1`. From August onward, use the current year.

Important: College vs. Pro Differences

• **Standings are per-conference** — use the `group` parameter to filter
• **Rankings replace leaders** — college uses AP Top 25, Coaches Poll, and CFP rankings
• **Ranked teams** have a `rank` field (null = unranked) on scoreboard competitors
• **Week-based schedule** — like NFL, college