Use this skill whenever the user asks about live sports scores, standings, team stats, game summaries (with box score, leaders, scoring plays, odds, and win probability), NFL / NBA / MLB / NHL / NCAA / MLS / EPL / WNBA games, team schedules, polls, or rankings. ESPN sports CLI with live scores across 10 leagues, offline search, head-to-head comparisons, and rich per-game summary payloads. No API key required. Triggers on natural phrasings like 'what's the score of the Lakers game', 'Patriots schedule this week', 'NFL standings', 'box score for tonight's Mavs game', 'Chiefs vs Eagles head to head', 'who's on top of the AP poll'.
Use this skill whenever the user asks about live sports scores, standings, team stats, game summaries (with box score, leaders, scoring plays, odds, and win probability), NFL / NBA / MLB / NHL / NCAA / MLS / EPL / WNBA games, team schedules, polls, or rankings. ESPN sports CLI with live scores across 10 leagues, offline search, head-to-head comparisons, and rich per-game summary payloads. No API key required. Triggers on natural phrasings like 'what's the score of the Lakers game', 'Patriots schedule this week', 'NFL standings', 'box score for tonight's Mavs game', 'Chiefs vs Eagles head to head', 'who's on top of the AP poll'.
This skill drives the espn-pp-cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first:
Install via the Printing Press installer. It defaults binaries to $HOME/.local/bin on macOS/Linux and %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows:
Ensure the reported install directory is on $PATH for the agent/runtime that will invoke this skill.
If the npx install fails (no Node, offline, etc.), fall back to a direct Go install (requires Go 1.26.5 or newer):
go install github.com/mvanhorn/printing-press-library/library/media-and-entertainment/espn/cmd/espn-pp-cli@latest
If --version reports "command not found" after install, the runtime cannot see the binary directory on $PATH. Do not proceed with skill commands until verification succeeds.
When to Use This CLI
Reach for this when a user wants a quick sports lookup - current score, standings, upcoming schedule, head-to-head record, or a rich per-game summary (box score, leaders, scoring plays, odds, win probability). Also good for cross-league discovery (today) and offline search across synced data.
Don't reach for this if the user has a paid feed like Stats Perform or Sportradar that provides cleaner data, or if they need real-time websocket updates (ESPN's endpoints are polling-only). For betting odds in isolation, the per-game summary payload includes them but there is no league-wide odds command.
Unique Capabilities
Commands that only work because of local sync + cross-league tooling.
Cross-league discovery
today — Today's scores across all major sports in one call. The fastest "what's on tonight" answer without picking a sport first.
trending — Most-followed athletes and teams across all leagues, ranked by current popularity. Good for "who is hot right now" without naming a sport.
dashboard — Reads [favorites] from ~/.config/espn-pp-cli/config.toml and shows scores for each favorited team across leagues, in one call.
watch <sport> <league> --event <game_id> — Live score updates for a specific game (polls every 30s). Use scores or today to find the game, then watch to follow it live.
Game-state intelligence
summary <sport> <league> --event <game_id> — Detailed game summary including box score, leaders, scoring plays, odds, and win probability. The single richest payload per game.
boxscore <event_id> — Just the per-player box score for an event id, with sport+league inferred from a recent scoreboard cache hit. Pass --sport/--league to skip inference.
plays <sport> <league> --event <id> — Play-by-play feed for a specific event. Optional --limit (default 200).
recap <sport> <league> — Post-game recap with box score and leaders for the most recent completed game in a league.
scoreboard <sport> <league> — Live scoreboard with date filtering, week/group selectors, and competition metadata.
odds <sport> <league> — Spread, over/under, and moneyline lines for tonight's slate, derived from the scoreboard payload (no per-game summary calls).
espn-pp-cli rankings <sport> <league> — AP, Coaches, and CFP polls
Info:
espn-pp-cli news <sport> <league> — Latest news
Discovery and local:
espn-pp-cli search "<query>" — Full-text search across synced events and news
espn-pp-cli sync — Sync a sport+league into local SQLite
espn-pp-cli sql "<query>" — Run read-only SQL against the local store
espn-pp-cli load — Show workload distribution per assignee (synced data)
espn-pp-cli orphans / stale — Maintenance views over the local store
espn-pp-cli doctor — Verify connectivity and configuration
Sport values: football, basketball, baseball, hockey, soccer.
League values: nfl, nba, mlb, nhl, ncaaf, ncaam, ncaaw, mls, eng.1 (EPL), wnba.
Recipes
Morning sports scan
espn-pp-cli today --agent --select events.shortName,events.status
espn-pp-cli scores football nfl --agent --select events.shortName,events.competitions.competitors.team.displayName,events.status.type.detail
espn-pp-cli standings football nfl --agent
One today call covers cross-league activity, one scores for the league you care about, one standings for context. The nested --select paths cut a scoreboard payload from tens of KB down to the fields that actually matter — essential for keeping agent context small.
Pre-game research from synced data
espn-pp-cli sync --sport football --league nfl
espn-pp-cli rivals football nfl --agent # historical records from synced data
espn-pp-cli streak football nfl --agent # current streaks
espn-pp-cli summary football nfl --event <id> --agent # full game payload incl. odds and box score
Run sync once, then rivals and streak answer instantly from the local store. summary is the richest single payload for a specific game (box score, leaders, scoring plays, odds, win probability).
Offline search after sync
espn-pp-cli sync --sport football --league nfl
espn-pp-cli search "Mahomes"# finds in local store
Useful for repeated lookups in poor-connectivity environments or when batch-analyzing historical data.
Favorites dashboard
Add a [favorites] block to ~/.config/espn-pp-cli/config.toml:
[favorites]
nfl = ["KC", "BAL"]
nba = ["LAL"]
Then:
espn-pp-cli dashboard --agent
One call surfaces tonight's matchup status for every favorited team, grouped by league. Per-league fetches run in parallel and partial failures are reported alongside successful results.
Pre-game odds and player digging
espn-pp-cli odds basketball nba --agent # tonight's spreads / totals / moneylines
espn-pp-cli leaders basketball nba --category points --agent
espn-pp-cli compare "LeBron James""Stephen Curry" --sport basketball --league nba --agent
espn-pp-cli boxscore <event_id> --agent # post-game player stats
espn-pp-cli plays basketball nba --event <id> --limit 50 --agent
odds reads the scoreboard's per-event lines (no per-game summary calls). leaders --category filters to one stat category. compare resolves athlete ids by name, listing candidates and exiting 2 on ambiguity. boxscore infers sport+league from the most recent cache hit; pass --sport/--league to skip inference.
Auth Setup
None required. ESPN's public endpoints don't require an API key. The auth command exists for consistency but is a no-op.
Optional config:
ESPN_CONFIG — override config file path
ESPN_BASE_URL — override base URL (for proxies or mirrors)
NO_COLOR — standard no-color env var
Agent Mode
Add --agent to any command. Expands to --json --compact --no-input --no-color --yes. Use --select for field cherry-picking, --dry-run to preview requests, --no-cache to bypass GET cache.
Filtering output
--select accepts dotted paths to descend into nested responses; arrays traverse element-wise:
Use this to narrow huge payloads to the fields you actually need — critical for deeply nested API responses.
Response envelope
Data-layer commands wrap output in {"meta": {...}, "results": <data>}. Parse .results for data and .meta.source to know whether it's live or local. The N results (live) summary is printed to stderr only when stdout is a TTY; piped/agent consumers see pure JSON on stdout.
Exit Codes
Code
Meaning
0
Success
2
Usage error
3
Not found (team, game, athlete)
5
API error
7
Rate limited
Installation
go install github.com/mvanhorn/printing-press-library/library/media-and-entertainment/espn/cmd/espn-pp-cli@latest
espn-pp-cli doctor
MCP Server
go install github.com/mvanhorn/printing-press-library/library/media-and-entertainment/espn/cmd/espn-pp-mcp@latest
claude mcp add espn-pp-mcp -- espn-pp-mcp
Argument Parsing
Given $ARGUMENTS:
Empty, help, or --help → run espn-pp-cli --help
install → CLI; install mcp → MCP
Anything else → resolve <sport> <league> from user intent (e.g., "Lakers" → basketball nba), check which espn-pp-cli (offer install if missing), run with --agent.
Automatic learning
Two-call protocol: recall before discovery, teach & before emitting. The CLI does entity-aware match validation AND surfaces stored playbooks for the query family; you read the envelope and follow the six-branch decision tree. Skipping either side costs you free recall hits in future sessions.
Step 1: recall before any discovery
Before scoreboard, teams, boxscore, search, standings, or any other discovery command on a new user question, run:
espn-pp-cli recall "<user's question>" --agent
The response envelope:
{"query":"...","normalized":"game tonight","query_entities":["Niners"],"found":true | false,"match_score":0.0,"results":[{"resource_id":"...","resource_type":"events|news|...","venue":"...","confidence":2,"entity_match":"exact|partial|unknown","source":"taught|preseed|pattern","warnings":["..."]}],"mismatches":[/* only when --debug-mismatches */],"warnings":[/* top-level */],"playbook":{"query_family":"...","playbook":{"steps":[{"cmd":"teams basketball nba {team.id}","purpose":"..."}, ... ],"entity_slots":["$TEAM","$STATS"],"expected_tool_calls":3},"slots_resolved":{"$TEAM":{"token":"pistons","canonical":"Detroit Pistons"}},"notes":"byathlete needs seasontype=2; categories has dup labels"},"notes":"byathlete needs seasontype=2; categories has dup labels"}
Step 2: six-branch decision tree
Read playbook, notes, results[0], and warnings in that order:
if Playbook present:
-> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose)
-> replay Playbook.steps in order, substituting Playbook.slots_resolved entries
for the entity slot tokens. If a step's slot is unresolved, fall back to
discovery for that step only.
-> the Playbook's expected_tool_calls is a budget; if you find yourself running
materially more, record the divergence via teach-playbook at end-of-session.
elif Notes present (no Playbook):
-> read Notes verbatim before any discovery step; they carry known gotchas
for this query family even when no structured choreography exists yet.
elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2:
-> skip discovery; fetch live data for Results[*].ResourceID in parallel
(e.g., espn-pp-cli boxscore <eid> for an event, espn-pp-cli teams ... for a team)
elif Found AND Results[0].EntityMatch == "partial":
-> candidate hint, NOT a hit; read the resource title to validate before trusting
elif (any row in Mismatches[] when --debug-mismatches was passed):
-> treat as cold start; the stored learning is for a different entity
(e.g., a "Cowboys" learning won't satisfy a "Niners" query — different canonical)
else: // Found == false, no playbook, no notes
-> cold start; run discovery normally; teach the answer afterward AND record
a playbook + notes via teach --playbook-file --playbook-notes-file so the
next session of the same family is faster.
Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a Results[] hit — use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping mismatches; pass --debug-mismatches only when investigating cold-start surprises.
Step 3: always read warnings
low_confidence: row exists at confidence<2. Treat as a hint, not a skip-discovery hit.
resource_not_in_store: the local store doesn't have the resource the learning points at. The match validator couldn't classify entities — direct-fetch and re-evaluate.
Top-level no_learnings_for_query_family: the table had no rows above the Jaccard floor. Pure cold start.
Step 4: teach & after finalizing your response
After assembling your user-facing response containing event/team IDs but BEFORE emitting it, fire the teach call in the background. Append a shell & so the call returns immediately and does not block the user-facing response:
Silent on success. Teach the most specific resource — if the user asked "when's next Spurs game" and you walked the teams scoreboard to find event 401747632, teach that event ID, not the team ID. The CLI uses the seeded entity_lookups (NFL/NBA/MLB/MLS team rosters with aliases like Niners/49ers/SF) for cross-alias resolution at recall time, so a teach under "Niners" will satisfy future "49ers" queries automatically.
Step 5: record a playbook when discovery took >5 calls
If you ran more than 5 tool calls to answer the question AND you discovered workarounds the CLI didn't surface (silently-dropped flags, undocumented params, pagination tricks, payload gotchas), record them so the next session of the same query family skips that discovery. Two surfaces:
# Common case: along with the resource teach
espn-pp-cli teach --query "<question>" --resource-type <type> --resource <id> \
--playbook-file ~/playbooks/season-recap.json \
--playbook-notes-file ~/playbooks/season-recap-notes.md
# (append shell `&` to background it)# Standalone: notes-only or recipe-only
espn-pp-cli teach-playbook --query "<question>" \
--playbook-file ~/playbooks/season-recap.json \
--notes "byathlete needs seasontype=2; categories has dup labels"
Playbook files are JSON with steps, entity_slots, expected_tool_calls. Notes files are markdown carrying the gotchas verbatim. Either field may be empty; both empty is rejected. Playbooks are keyed on the structural query family (entities stripped) so a recipe taught from "how did Warriors end the season" applies to Pistons, Heat, and every other team query of the same shape.
When you DO find a playbook on a future recall, treat it as ground truth: replay the steps with slots_resolved substitutions, skip the discovery that the choreography already documents, and read notes before any step.
Step 6: playbook amend & when your debug response identifies a correction
If your debug-protocol response identifies a concrete correction the notes or playbook should know — a workaround, an undocumented endpoint shape, a stale field name, observed schema drift, an empty-payload fallback — fire playbook amend BEFORE emitting your user-facing response. Same fire-and-forget posture as teach.
What counts as worth amending: a behavior you OBSERVED this session that future-you would benefit from knowing. Examples worth amending:
"summary endpoint wraps response in {meta, results} -- data lives at .results.header, not .header"
"compare returns empty team field for athletes whose team isn't in current playoffs"
"byathlete category index for avgRebounds drifted from general[9] to general[11] this season"
What does NOT belong in notes:
The year-specific answer ("Curry shot 4-16 in his last game"). That's the response, not a learning.
Per-team or per-athlete data that the playbook already retrieves at runtime.
Statements that paraphrase what the existing notes already say.
The amend command appends to the family's existing notes with a timestamped marker ([amend YYYY-MM-DDTHH:MMZ]: <text>). Multiple amends accumulate; the audit trail is visible. If no playbook exists yet for the family, amend creates a notes-only one (so cold-start corrections still land).
Worked examples
Cold: "when's next Spurs game?" — recall returns found=false. Walk the teams scoreboard, find next upcoming Spurs event ID, answer. Teach the event:
espn-pp-cli recall "when's next Spurs game?" --agent
# found=false -> discovery
espn-pp-cli teams basketball nba 24 --agent --select events.id,events.shortName,events.date
# ...answer "Tue May 26 SA @ OKC, event 401747632"...
espn-pp-cli teach --query "when's next Spurs game?" --resource-type events --resource 401747632
# (append shell `&` to background it)
Warm: "Niners game tonight" — recall returns found=true, results[0].entity_match="exact", results[0].confidence>=2. Skip discovery; fetch the box score directly:
Cross-alias hit: "49ers game tonight" — never directly taught. recall resolves "49ers" → "San Francisco 49ers" canonical via entity_lookups (nfl_team kind), finds the "Niners game tonight" learning (same canonical), returns found=true. Skip discovery.
Entity mismatch: "Cowboys game tonight" — has a Niners learning above the Jaccard floor on non-entity tokens (game, tonight), but the entity canonical differs (Dallas Cowboys ≠ San Francisco 49ers). Filtered into mismatches; recall returns found=false. Treat as cold start.
When the loop is broken: learnings list --warnings surfaces local issues; espn-pp-cli feedback "<what tripped you up>" records the friction so the next print can fix it.
Agent Workflow Features
This CLI exposes three shared agent-workflow capabilities patched in from cli-printing-press PR #218.
Named profiles
Persist a set of flags under a name and reuse them across invocations.
# Save the current non-default flags as a named profile
espn-pp-cli profile save <name>
# Use a profile — overlays its values onto any flag you don't set explicitly
espn-pp-cli --profile <name> <command>
# List / inspect / remove
espn-pp-cli profile list
espn-pp-cli profile show <name>
espn-pp-cli profile delete <name> --yes
Flag precedence: explicit flag > env var > profile > default.
--deliver
Route command output to a sink other than stdout. Useful when an agent needs to hand a result to a file, a webhook, or another process without plumbing.
File sinks write atomically (tmp + rename). Webhook sinks POST application/json (or application/x-ndjson when --compact is set). Unknown schemes produce a structured refusal listing the supported set.
feedback
Record in-band feedback about this CLI from the agent side of the loop. Local-only by default; safe to call without configuration.
espn-pp-cli feedback "what surprised you or tripped you up"
espn-pp-cli feedback list # show local entries
espn-pp-cli feedback clear --yes# wipe
Entries append to ~/.espn-pp-cli/feedback.jsonl as JSON lines. When ESPN_FEEDBACK_ENDPOINT is set and either --send is passed or ESPN_FEEDBACK_AUTO_SEND=true, the entry is also POSTed upstream (non-blocking — local write always succeeds).
