Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Cricket data via ESPN public endpoints and Cricsheet open data — live-ish series scoreboards, standings, match summaries and news (ESPN), plus historical ball-by-ball, player stats, and player registry (Cricsheet, ODC-BY 1.0). Zero config, no API keys. Use when: user asks about cricket scores, IPL/BBL/PSL/international series, points tables, match details, cricket news, ball-by-ball history, or player career stats. Don't use when: user asks about other sports — use football-data (soccer), nfl-d
.claude/skills/machina-sports-cricket-data/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 192% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 528% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 94% | 0% |
Before writing queries, consult references/api-reference.md for endpoints, ID conventions, and data shapes. See references/competitions.md for the Cricsheet competition codes.
Prefer the CLI — it avoids Python import path issues. There are two backends: ESPN (live-ish) and Cricsheet (historical).
bash# ESPN backend (live-ish) — discover series first, then use its numeric ID sports-skills cricket get_series sports-skills cricket get_scoreboard --series_id=8048 sports-skills cricket get_standings --series_id=8048 sports-skills cricket get_game_summary --series_id=8048 --event_id=1535465 sports-skills cricket get_news --series_id=8048 # Cricsheet backend (historical, completed matches) — uses letter codes sports-skills cricket get_competitions sports-skills cricket get_matches --competition=ipl --season=2026 sports-skills cricket get_match_deliveries --competition=ipl --match_id=1473508 sports-skills cricket get_player_stats --competition=ipl --player="V Kohli" sports-skills cricket find_player --name=kohli
CRITICAL: Before calling any data endpoint, verify:
get_series first. IDs change every season for recurring tournaments (e.g. each IPL season has a different ID). Never hardcode them — the 8048 in the examples is illustrative, not permanent.get_series returns ESPN IDs (live, numeric, e.g. 8048). get_competitions returns Cricsheet codes (historical, letter codes, e.g. ipl). They are unrelated except at the match level: a Cricsheet match_id equals the ESPNcricinfo match ID, so it bridges the two backends.get_scoreboard, get_series).get_player_stats requires the exact Cricsheet name spelling (e.g. "V Kohli", not "Virat Kohli"). Resolve the spelling with find_player first.get_standings, which is empty for most bilateral tours (only league/group tournaments publish a points table).~/.cache/sports-skills/cricsheet/; later calls in the same day are fast.Agents can run scripts/validate_params.sh to pre-validate --competition, --series_id, and --date before querying.
| Command | Required params | Description | |---|---|---| | get_series | — | List currently-active cricket series with ESPN series IDs and live events | | get_scoreboard | series_id (opt date) | Matches + scores + status for a series | | get_standings | series_id | Points table for a series (empty for most bilateral tours) | | get_game_summary | series_id, event_id | Match detail: rosters, leaders, matchcards, venue info | | get_news | series_id | News articles for a series |
| Command | Required params | Description | |---|---|---| | get_competitions | — | List Cricsheet competition codes | | get_matches | competition (opt season) | Completed matches for a competition, newest first | | get_match_deliveries | competition, match_id (opt innings) | Ball-by-ball deliveries for a completed match | | get_player_stats | competition, player (opt season) | Aggregate batting + bowling stats for a player | | find_player | name | Search Cricsheet player registry; returns ESPNcricinfo ID mappings |
Dates accept YYYYMMDD or YYYY-MM-DD. season is the start year (e.g. 2020 matches Cricsheet's "2020/21"). See references/api-reference.md for full parameter lists and return shapes.
get_series → pick the series and note its series_id.get_scoreboard --series_id=<id> → present matches by status and score.get_game_summary --series_id=<id> --event_id=<id>.get_series → find the tournament's series_id.get_standings --series_id=<id>. If empty, it is a bilateral tour with no published table.find_player --name=<substring> → confirm the exact name spelling.get_player_stats --competition=<code> --player="<exact name>" (optionally --season).get_matches --competition=<code> [--season=<year>] → find the match_id.get_match_deliveries --competition=<code> --match_id=<id> [--innings=N].Cricket news is series-scoped — there is no global feed.
get_series → pick the relevant series and note its series_id.get_news --series_id=<id> → present the articles.get_player~~ / ~~get_player_info~~ — do not exist. Use find_player to resolve a name, then get_player_stats for career numbers.get_match~~ — does not exist. Use get_matches (historical list) or get_game_summary (one ESPN match's detail).get_rankings~~ — does not exist. There is no free ICC rankings source (v1 limitation).get_teams~~ / ~~get_team_roster~~ — do not exist. Rosters come inside get_game_summary.get_play_by_play~~ — does not exist. Use get_match_deliveries for ball-by-ball data on completed matches.If a command is not listed in the Commands table above, it does not exist.
Cricsheet data is licensed ODC-BY 1.0. Every Cricsheet response includes an attribution field ("Data from Cricsheet (cricsheet.org), ODC-BY 1.0"). Preserve this attribution string when republishing the data.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | fail→fail | 3,716 | 5,207 | +40% | 1 | 1 | 0% | 614 | 1,940 | +216% | 0 | 0 | — |
case-04 | fail→pass | 4,552 | 3,699 | -19% | 1 | 1 | 0% | 788 | 2,303 | +192% | 0 | 0 | — |
case-01 | fail→fail | 8,126 | 2,351 | -71% | 1 | 1 | 0% | 1,518 | 1,838 | +21% | 0 | 0 | — |
case-02 | fail→fail | 6,950 | 4,258 | -39% | 1 | 1 | 0% | 1,207 | 1,880 | +56% | 0 | 0 | — |
case-03 | fail→fail | 17,420 | 5,339 | -69% | 1 | 1 | 0% | 3,746 | 1,926 | -49% | 0 | 0 | — |
case-15 | pass→pass | 7,726 | 2,802 | -64% | 1 | 1 | 0% | 1,410 | 2,022 | +43% | 0 | 0 | — |
case-05 | fail→fail | 6,195 | 12,805 | +107% | 1 | 1 | 0% | 1,106 | 3,461 | +213% | 0 | 0 | — |
case-06 | pass→fail | 2,986 | 4,272 | +43% | 1 | 1 | 0% | 477 | 2,048 | +329% | 0 | 0 | — |
case-07 | fail→fail | 4,692 | 7,686 | +64% | 1 | 1 | 0% | 751 | 2,158 | +187% | 0 | 0 | — |
case-08 | fail→fail | 6,533 | 8,950 | +37% | 1 | 1 | 0% | 1,220 | 2,427 | +99% | 0 | 0 | — |
case-09 | fail→fail | 14,156 | 5,496 | -61% | 1 | 1 | 0% | 2,361 | 1,972 | -16% | 0 | 0 | — |
case-10 | fail→fail | 7,756 | 4,754 | -39% | 1 | 1 | 0% | 1,319 | 1,984 | +50% | 0 | 0 | — |
case-11 | pass→fail | 6,294 | 5,708 | -9% | 1 | 1 | 0% | 1,003 | 2,080 | +107% | 0 | 0 | — |
case-12 | fail→pass | 3,821 | 10,465 | +174% | 1 | 1 | 0% | 491 | 3,084 | +528% | 0 | 0 | — |
case-13 | fail→pass | 9,174 | 2,451 | -73% | 1 | 1 | 0% | 1,774 | 2,105 | +19% | 0 | 0 | — |
case-14 | pass→pass | 11,078 | 6,950 | -37% | 1 | 1 | 0% | 1,906 | 2,792 | +46% | 0 | 0 | — |
case-16 | pass→pass | 8,026 | 4,146 | -48% | 1 | 1 | 0% | 1,531 | 2,365 | +54% | 0 | 0 | — |
case-17 | fail→pass | 8,730 | 1,660 | -81% | 1 | 1 | 0% | 1,582 | 1,977 | +25% | 0 | 0 | — |
case-18 | fail→fail | 8,276 | 10,573 | +28% | 1 | 1 | 0% | 1,379 | 2,373 | +72% | 0 | 0 | — |
case-19 | fail→fail | 6,198 | 7,907 | +28% | 1 | 1 | 0% | 623 | 2,184 | +251% | 0 | 0 | — |
case-20 | fail→pass | 5,408 | 1,580 | -71% | 1 | 1 | 0% | 961 | 1,860 | +94% | 0 | 0 | — |
case-22 | fail→fail | 3,427 | 6,193 | +81% | 1 | 1 | 0% | 574 | 2,098 | +266% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 9 counted toward the lift figure. The other 13 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +14 percentage points is the difference between those two pass rates over the 9 comparable cases. 5 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.