Audit websites for accessibility issues and WCAG compliance. Use when checking accessibility, fixing a11y issues, or ensuring WCAG compliance.
日本語の概要は準備中です。原文の説明を表示しています。
Runs small and midsize business bookkeeping operations - categorizes bank and card transactions to a chart of accounts, reconciles a bank or card statement against the ledger, works a month-end close checklist, and produces a close package for the accountant. Works from QuickBooks Online, Xero, bank CSV/OFX, or Mercury exports, and from the Xero, QuickBooks, or Mercury connectors when present. Use it whenever the user says "reconcile", "bank rec", "categorize transactions", "month end close", "close the books", "books don't match the bank", "reconciliation is off by", "uncategorized expenses", or shares a QuickBooks or Xero export or a bank statement CSV, even if they do not name the task. Never plugs a difference; every unreconciled dollar is explained or listed.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
A reconciliation proves that the books and the bank agree once timing is accounted for. Models get this wrong in predictable ways: they flip signs, treat an uncleared check as an error, count a transfer between the business's own accounts as income, miss a transaction imported twice, and force the rec to zero with an adjustment. This skill exists to prevent each of those. The deterministic matching lives in scripts/reconcile.py; the judgment lives here.
Scope: bookkeeping, not tax. When an item has a tax consequence (capitalize vs expense, owner draws vs wages, sales tax, 1099 status, meals deductibility), record the bookkeeping fact and put it on the accountant list. Do not give tax advice.
For each account being reconciled (checking, savings, each credit card, loan, payment processor clearing):
listTransactions, or the platform's statement lines.See references/platform-notes.md for export paths, column formats, and connector limits. Neither the QuickBooks Online API nor the Xero Accounting API can mark items reconciled; reconciliation is always finished inside the product.
Convert every file to one convention: amount is the signed change to the balance the statement reports. For a bank account, deposits positive. For a card, charges positive. Watch the word "debit": on a bank statement it means money out, in a general ledger it means money into a cash account. reconcile.py handles both and prints which reading it used. Confirm dates are parsed in the right day/month order before trusting anything.
Beginning balance plus statement lines must equal ending balance. If it does not, the export is incomplete, overlaps another period, or contains a duplicated download. Stop and fix the input. Matching against a broken statement produces a rec that balances to the wrong number.
Run:
python scripts/reconcile.py --bank bank.csv --ledger ledger.csv \
--period-start YYYY-MM-DD --period-end YYYY-MM-DD \
--statement-begin N --statement-end N --book-begin N --out recon/
It matches in tiers (exact, check number, same amount within a date window, near amount with similar description, one-to-many for batched deposits) and classifies everything left over. Read its notes: it reports when it inverted a file's signs. Exit 0 = balanced, 2 = unexplained variance, 3 = statement does not roll forward.
Interpret the results with these rules:
If the adjusted bank and adjusted book balances still differ, report the variance with what was ruled out. Do not create a "reconciliation discrepancy" or suspense entry to force zero. QuickBooks describes its adjusting entry as a last resort for small amounts only, and it hides the underlying error. If the user insists on an adjustment, list it as a proposed entry for accountant approval, with the amount and the investigation done.
Common causes to check, in order: wrong beginning balance (a previously reconciled item was edited or deleted), wrong statement balance typed in, a line outside the date range, a sign flip on a single line, a duplicate, a transfer recorded twice, a split transaction entered as the gross amount.
Apply references/categorization-rules.md: existing bank rules first, then vendor patterns, then history. Every categorization carries a confidence: high (rule or consistent history), medium (pattern match only), low (guess). Only high is posted without review. Items in Uncategorized, Ask My Accountant, or suspense are never left there at close.
Follow references/close-checklist.md. Short version: all accounts reconciled including cards, loans, and processor clearing; AR and AP aging tie to the balance sheet; payroll liabilities tie; undeposited funds cleared; accruals and prepaids posted; fixed asset additions flagged; balance sheet reviewed line by line; P&L compared to prior month and prior year; period locked only after the accountant signs off.
Produce one document per period, in this order:
reconciliation.md, balanced or not, with unexplained variance stated in dollars.FOR ACCOUNTANT REVIEW and never posted by this skill: date, debit account, credit account, amount, memo, evidence. Examples: record bank fee, remove duplicate, reclass transfer out of income, accrue unbilled expense.State plainly what was not verified (for example, "loan statement not provided, loan balance not reconciled").
See examples/worked-example.md for a full run on the bundled fixtures.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Audit websites for accessibility issues and WCAG compliance. Use when checking accessibility, fixing a11y issues, or ensuring WCAG compliance.
日本語の概要は準備中です。原文の説明を表示しています。
Deploy a 2-layer parallel agent hierarchy for large, parallelizable work — big refactors, multi-file migrations, codebase-wide audits, bulk generation. A top-tier commander (Fable or Opus) orchestrates the swarms; the user picks a power level (Max Power / Heavy / Balanced / Economy) that sets the Opus/Sonnet/Haiku model mix per layer. Layer 1 is 3-50+ specialist agents, each with its own full context window; Layer 2 is 2+ sub-agents per member. Includes git safety, tiered sizing, a pre-deploy gate, phantom-completion checks, and multi-wave follow-up.
日本語の概要は準備中です。原文の説明を表示しています。
Use to test whether an AI agent, prompt, or skill actually works before or after it ships — build a small eval set, score runs, and catch regressions when the prompt or model changes. Trigger when the user asks whether a prompt is good, wants to compare models or versions, sees inconsistent agent output, or is about to put an agent in front of customers.
日本語の概要は準備中です。原文の説明を表示しています。
Deploys swarms of sub-agents for massive parallel data processing tasks. Unlike agent-army (which is for code changes), this is for DATA tasks -- processing 1000 documents, analyzing datasets, bulk content generation. Configurable swarm size, task distribution, result aggregation, progress tracking, and error recovery.
日本語の概要は準備中です。原文の説明を表示しています。
Designs and deploys custom agent teams for specific business workflows. Interactive discovery of business processes, then generates complete team configurations with specialized agent roles, tool access, communication protocols, and handoff rules.
日本語の概要は準備中です。原文の説明を表示しています。
Agent-to-Agent (A2A) communication protocol. Connect two or more Claude agents that pass messages, share context, delegate tasks, and collaborate. Implements structured handoffs, shared memory, and multi-agent conversations.
日本語の概要は準備中です。原文の説明を表示しています。