skills/mlb-league-state-reader/SKILL.md
Parses Yahoo Fantasy Baseball league state (roster, standings, current matchup, FAAB remaining, free agents) from authenticated Yahoo team pages via Claude-in-Chrome browser automation, then grounds it against league-config.md and team-profile.md to emit a normalized league-state bundle every other agent can consume without re-scraping. Use when the coach or any downstream agent needs to read Yahoo roster, refresh team profile, pull league state, get current matchup, check FAAB remaining, list free agents, or when user mentions "what's on my roster", "who am I playing this week", "how much FAAB do I have left", or "refresh my team".
npx skillsauth add lyndonkl/claude mlb-league-state-readerInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Scenario: Morning brief run on 2026-04-17. Coach needs current roster, matchup, FAAB, and standings before launching lineup-optimizer.
Inputs read first:
yahoo-mlb/context/league-config.md (league ID 23756, team 5, 12 teams, 5x5 cats)yahoo-mlb/context/team-profile.md (prior known roster skeleton)Chrome navigation sequence (authenticated):
https://baseball.fantasysports.yahoo.com/b1/23756/5 — roster with today's opponents per playerhttps://baseball.fantasysports.yahoo.com/b1/23756?lhst=stand — standings + category rankshttps://baseball.fantasysports.yahoo.com/b1/23756/matchup — this week's head-to-head vs Los Doyershttps://baseball.fantasysports.yahoo.com/b1/23756/transactions — FAAB spent to date → remaininghttps://baseball.fantasysports.yahoo.com/b1/23756/players?status=A — top free agents (page 1)Extracted bundle:
| Field | Value | |---|---| | Record | 8-6-1 | | Rank | 5 of 12 | | FAAB spent | $14 | | FAAB remaining | $86 | | This week opp | Los Doyers | | Live matchup score | 4-3-3 (us leading R, HR, K; losing OBP, ERA, WHIP) | | Roster returned | 26 of 26 slots populated | | Free agents top-20 | captured for waiver-analyst |
Outputs written:
context/team-profile.md — roster snapshot updated in place (slot, player, MLB team, today's opp, status)signals/2026-04-17-league-state.md — signal file with YAML frontmatter, source_urls, confidenceEvery downstream agent (lineup-optimizer, waiver-analyst, category-strategist) now reads this signal file and team-profile.md rather than re-scraping Yahoo.
Copy this checklist and track progress:
League State Reader Progress:
- [ ] Step 1: Read context files (league-config.md + team-profile.md)
- [ ] Step 2: Pull Yahoo pages via Claude-in-Chrome (5 URLs)
- [ ] Step 3: Parse roster, standings, matchup, FAAB, free agents
- [ ] Step 4: Update team-profile.md in place
- [ ] Step 5: Emit signal file with frontmatter + source_urls
Step 1: Read context files
Load the authoritative contract and the last known state before touching Yahoo. This both verifies the league ID/URL and provides the skeleton to diff against.
yahoo-mlb/context/league-config.md — confirm league ID 23756, team 5, roster shape (C/1B/2B/3B/SS/3OF/2UTIL/3SP/2RP/5P/3BN/3IL = 26), FAAB budget $100yahoo-mlb/context/team-profile.md — capture prior FAAB remaining, prior roster for diffyahoo-mlb/context/frameworks/signal-framework.md — confirm signal file formatyahoo-mlb/context/frameworks/data-sources.md — confirm the 5 Yahoo URLsStep 2: Pull Yahoo pages via Claude-in-Chrome
The user is already authenticated in the browser. Navigate in order; parse page text; if any step fails, jump to Degraded Mode for that specific URL only (do not abort the whole run). See resources/methodology.md for the exact tool-call sequence and page-parsing patterns.
https://baseball.fantasysports.yahoo.com/b1/23756/5https://baseball.fantasysports.yahoo.com/b1/23756?lhst=standhttps://baseball.fantasysports.yahoo.com/b1/23756/matchuphttps://baseball.fantasysports.yahoo.com/b1/23756/transactionshttps://baseball.fantasysports.yahoo.com/b1/23756/players?status=AStep 3: Parse and normalize
Extract structured data from each page. See resources/methodology.md for row-level parsing logic, and resources/methodology.md for how to reconstruct FAAB remaining from the transactions log.
{slot, player, mlb_team, opp_today, status: active|IL|DTD|NA|BN}-$N entries from transactions history this season; remaining = $100 - total spentStep 4: Update team-profile.md in place
See resources/template.md for the exact output shape. Overwrite the existing file (not append); this file is meant to always reflect "as of the latest run."
## As of YYYY-MM-DDStep 5: Emit signal file
Write yahoo-mlb/signals/YYYY-MM-DD-league-state.md with YAML frontmatter per the signal framework. Every URL navigated must appear in source_urls. Degraded fetches drop confidence to 0.5; full-failure fields drop to 0.3 and are flagged. See resources/template.md.
type: league-state, date, emitted_by: mlb-league-state-reader, confidence, source_urls[]mlb-signal-emitter before writing (if that skill is available in this run); otherwise write directlyChrome can fail for several reasons: tab not open, session expired, Yahoo rate-limiting, 500 error, unreadable DOM. Never abort the whole run. Fail gracefully per-URL.
Per-URL fallback decision tree:
| Failure | Fallback |
|---|---|
| Navigation error / timeout | Retry once. If still failing, ask user: "Paste the contents of {URL} and I'll parse from that." |
| Session expired (login page returned) | Ask user: "Please log into Yahoo in your Chrome tab, then say 'retry'." Do not attempt to re-authenticate. |
| Page loads but parse fails (unknown layout) | Dump the first ~4000 chars of get_page_text into the signal file under raw_capture: and drop confidence to 0.4 for that field. |
| Free-agents page gated / empty | Mark free_agents: [] and flag in the signal. The waiver-analyst can still run on roster + standings. |
| Transactions page gated | Ask user: "What's your FAAB remaining right now? Visible at the top of the transactions page." Accept user value; mark faab_confidence: 0.6. |
User-paste template (reuse for any Yahoo page):
"I couldn't reach
{URL}. Please copy the visible table on that page (select all, paste into chat) and I'll extract the fields I need from your paste."
Parse the pasted text with the same regex/row patterns as the live page. See resources/methodology.md.
Never assume the roster is still what it was yesterday. Waivers clear daily and trades can fire any time. Always re-read the Yahoo team page at the start of every run, even if team-profile.md looks fresh.
FAAB remaining is derived, not displayed. Yahoo's UI sometimes shows "FAAB Balance" and sometimes doesn't. The authoritative computation is: $100 minus sum of all winning bids on the transactions page this season. Do not trust a single visible number over the ledger.
Cite every URL. The signal file's source_urls: list must contain every Yahoo page actually visited. If a page was skipped because of degradation, note it in the signal body, not silently.
Do not re-scrape downstream. Once this skill has emitted today's league-state signal, lineup-optimizer / waiver-analyst / category-strategist MUST read that signal rather than re-visit Yahoo. Re-scraping wastes tokens and risks drift.
Overwrite team-profile.md; append nowhere. This file is "current state." Preserving yesterday's roster pollutes it. The append-only log lives in tracker/decisions-log.md, not here.
Status values are normalized enums. Map Yahoo's labels (P, DTD, IL-10, IL-60, NA, SUSP) to the normalized enum: active | BN | IL | DTD | NA | SUSP. Downstream agents branch on the enum, not Yahoo's strings.
Today's opponent may be blank. Off-days, postponements, and doubleheaders all produce unusual opp_today values. Record exactly what Yahoo shows (@TEX, TEX, Off, PPD, DH1/DH2) — do not normalize away information the downstream lineup-optimizer needs.
Don't emit a signal with zero source URLs. If every page failed, write an empty-body signal with confidence: 0.0 and a prominent note to the coach: "Could not reach Yahoo — paste or retry required before any downstream agent runs."
The five Yahoo URLs (from context/frameworks/data-sources.md):
Team: https://baseball.fantasysports.yahoo.com/b1/23756/5
Standings: https://baseball.fantasysports.yahoo.com/b1/23756?lhst=stand
Matchup: https://baseball.fantasysports.yahoo.com/b1/23756/matchup
Transactions: https://baseball.fantasysports.yahoo.com/b1/23756/transactions
Free agents: https://baseball.fantasysports.yahoo.com/b1/23756/players?status=A
Chrome tool order (load via ToolSearch first, then call):
1. mcp__claude-in-chrome__tabs_context_mcp (confirm active tab, auth state)
2. mcp__claude-in-chrome__navigate (go to URL)
3. mcp__claude-in-chrome__get_page_text (extract DOM text)
4. mcp__claude-in-chrome__read_page (structured read if text fails)
Roster slot order (for team-profile YAML; matches league-config.md):
C, 1B, 2B, 3B, SS, OF, OF, OF, UTIL, UTIL, SP, SP, SP, RP, RP, P, P, P, P, P, BN, BN, BN, IL, IL, IL
Status enum mapping:
| Yahoo | Normalized |
|---|---|
| (blank) / P | active |
| BN | BN |
| DTD | DTD |
| IL-10 / IL-60 / IL | IL |
| NA | NA |
| SUSP | SUSP |
Signal frontmatter (minimum fields):
---
type: league-state
date: 2026-04-17
emitted_by: mlb-league-state-reader
variant_synthesis: false
confidence: 0.9
source_urls:
- https://baseball.fantasysports.yahoo.com/b1/23756/5
- https://baseball.fantasysports.yahoo.com/b1/23756?lhst=stand
- https://baseball.fantasysports.yahoo.com/b1/23756/matchup
- https://baseball.fantasysports.yahoo.com/b1/23756/transactions
- https://baseball.fantasysports.yahoo.com/b1/23756/players?status=A
---
Key resources:
Inputs required:
context/league-config.md (league ID, team ID, roster shape)context/team-profile.md (prior state, may be stub)Outputs produced:
context/team-profile.md (overwrite; reflects today's state)signals/YYYY-MM-DD-league-state.md (with frontmatter + source_urls)testing
Cluster a conference's event records into a small set of coarse themes with finer sub-clusters, an explicit outlier bucket, and soft (multi-membership) affinities — using the hybrid embed-then-label pipeline (embed abstracts, reduce, density-cluster, then LLM-label the clusters) when embedding libraries are available, and an LLM-reasoned hierarchical fallback when they are not. Embeddings do the grouping; the LLM only names the groups. Conference-agnostic. Use when turning structured event records into a navigable theme map for preference elicitation and scheduling, when you need 6-8 reasonable themes rather than 20 muddy ones, or when overlapping talks must belong to more than one theme. Trigger keywords - theme clustering, cluster talks, embed then label, soft membership, outlier talks, conference themes, topic map.
development
Build a personal conference schedule as a constraint-optimization problem — hard constraints (no time overlap, room-to-room travel time, capacity/registration, the attendee's own must-attends and blackouts) plus a user-owned weighted objective trading interest against breadth, pacing (maximize contiguous free time), and serendipity. Surfaces unbreakable conflicts (two high-value overlapping talks the model cannot rank) as decisions for the human rather than silently picking, and reports what each choice traded away. Conference-agnostic. Use to turn a preference profile plus a theme map into a day-by-day plan, to resolve overlapping sessions, or to balance a packed vs paced schedule. Trigger keywords - schedule optimization, conference schedule, constraint optimization, overlapping talks, contiguous free time, conflict surfacing, packed vs paced.
development
Parse a heterogeneous conference program (markdown, HTML, PDF-derived text, or JSON) into normalized event records with per-field confidence scores and independent classification axes (topic, depth, format, prerequisites, recorded, capacity). Detects the program's format before extracting, treats every inferred field as uncertain (present vs inferred vs missing), and flags thin or missing abstracts so downstream enrichment can target them. Conference-agnostic. Use when ingesting a conference or event schedule into a structured store, normalizing a talk/session list, or extracting per-session metadata with calibrated confidence. Trigger keywords - program ingestion, parse schedule, session extraction, event records, conference program, talk metadata, per-field confidence.
development
Build a personalized preference profile from a small number of well-chosen, cluster-grounded questions instead of a long survey. Represents the person's interests as an uncertainty region over the theme map, picks the single highest-information-gain choice-based question (contrasting real talks from different clusters), balances exploiting known interests against exploring uncertain ones, deliberately injects outlier probes to fight selection bias, and stops as soon as the schedule would be stable. Also elicits the user-owned objective weights and hard constraints. Interactive — runs where it can actually ask the person. Conference-agnostic. Use to turn a theme map into a preference profile, to decide what to ask a conference attendee, or to elicit scheduling priorities. Trigger keywords - preference elicitation, ask few questions, information gain, choice-based questions, selection bias probe, objective weights, attendee preferences.