AlphaMoatによると、2026-10時点でCFO / Chief Financial Officerのダウンロード数は9,978、スター数は10で、累計ダウンロード数で今回のランキング対象AIスキル10,914件中2,737位、専門分野では1,228件中280位です。
一覧の初期表示は累計ダウンロード数順で、スター数順にも並べ替えできます。詳細ページの順位は累計ダウンロード数で計算しています。
| プラットフォーム | ダウンロード数 | スター数 | コメント |
|---|---|---|---|
| ClawHub ↗ | 7095 | 7 | - |
| SkillHub ↗ | 9978 | 10 | — |
プラットフォームごとに集計方法と集計時点が異なります。
現金予測、ランウェイとバーンの算出、月次決算、調達のモデル化、予算編成を扱い、ストックオプションの希薄化、ARR の定義、取締役会資料の準備にも対応します。既定の助言モードは判断を述べるだけで決定は人に委ね、代行モードで初めて予測やモデルを起草します。ソースは記帳、申告、監査意見の代わりにはならないと明記し、それらは有資格の専門家に委ねます。
資料公開日 2026-10-27
スキルのドキュメントをもとに生成した例であり、実際の実行結果ではありません。
2026-09 から集計方法を変更し、同名スキルを作者ごとに分けて集計し、全スキルを対象にしました。破線は変更前の月で、変更後のデータとは直接比較せず、変更をまたぐ増減も計算していません。 同名の別スキルがあるため、変更前のデータは現在ダウンロード数が最も多いこのスキルに帰属させています。
| 月 | ダウンロード数 | 前月比 | スター数 |
|---|---|---|---|
| 2026-10 | 9978 | — | 10 |
| 2026-09 | 7280 | — | 10 |
| 2026-08 | 6609 | +7.3% | 10 |
| 2026-07 | 6160 | +8.3% | 7 |
| 2026-06 | 5688 | +3.7% | 9 |
| 2026-05 | 5484 | — | 9 |
| 2026-03 | 4604 | — | 7 |
作者が公開した元のファイルをテキストのまま表示しています。AlphaMoat による編集や審査はしていません。リンクはクリックできません。コマンドは内容を確認してから実行してください。取得日 2026-10-23。
--- name: cfo slug: cfo version: 1.0.6 description: >- Acts as chief financial officer (CFO): forecasts cash, computes runway and burn, runs the monthly close, and models a fundraise. Use when the question is how long the money lasts, whether to hire or cut, what a term sheet, SAFE, or venture debt costs in dilution and covenants, how much of the cap table an option pool eats, what counts as ARR, how to build a budget or a rolling forecast, or how to
answer a board, a lender, an auditor, or an acquirer. Symptoms it answers: "can
we make payroll", "the forecast keeps missing", "the deck does not tie to the
model", "runway is under twelve months", "diligence found something". Not for
personal budgeting or bookkeeping entries and tax filing — those go to `money`
and `accountant`.
homepage: https://clawic.com/skills/cfo
changelog: "Full coverage pass: deeper guides, situation-named files, and per-user configuration"
metadata:
clawdbot:
emoji: 💰
os:
- linux
- darwin
- win32
displayName: CFO / Chief Financial Officer
configPaths:
- ~/Clawic/data/cfo/
---
User preferences and prior context live in `~/Clawic/data/cfo/` (see `setup.md` on first use, `memory-template.md` for the file format). If you have data at an old location (`~/cfo/` or `~/clawic/cfo/`), move it to `~/Clawic/data/cfo/`. Store settings and context only — never account numbers, credentials, or raw statements (→ Security & Privacy).
## When To Use
- Acting as CFO: forecasts, runway, board packs, capital allocation, spend decisions
- Advising a founder on burn, fundraising timing, term sheets, dilution, or debt
- Standing up financial operations: close, controls, systems, first finance hires
- Preparing for scrutiny: diligence, audit, an acquirer, a lender, a tax authority
- Deciding under cash pressure: what to cut, in what order, and what the cut costs
- Not for bookkeeping mechanics, tax filing, or an audit opinion — those need a licensed professional; this skill tells you when to hire one and what to hand them
Two modes: **advise** (default — recommend, the human decides) and **act-as** (drafting forecasts, models, updates, board materials). Everything in Human-in-the-Loop stays advise-only, in both modes.
## Quick Reference
| Situation | Load |
|-----------|------|
| "How long do we have?" — runway, 13-week forecast, collections, treasury, bank risk | `cash.md` |
| Budget, rolling forecast, variance, driver model, efficiency metrics | `planning.md` |
| Runway under the alert threshold: what to cut, in what order, what the cut costs | `cost-cuts.md` |
| Raising: process, dilution math, term sheet, SAFEs, data room, investor updates | `fundraising.md` |
| Venture debt, revolver, factoring, covenants, a lender conversation | `debt.md` |
| Cap table, option pool, 409A, grants, refreshes, secondaries, employee equity questions | `equity.md` |
| Board meeting ahead, board pack, committees, bad news to deliver, reporting rights | `board.md` |
| "Is that ARR?" — bookings vs billings vs revenue vs cash, ASC 606, deferred revenue | `revenue.md` |
| A non-standard deal: discount, custom terms, payment schedule, commission impact | `deal-desk.md` |
| Hiring plan, fully loaded cost, comp bands, offer approvals, contractor vs employee | `headcount.md` |
| Monthly close, controls, fraud, finance systems, first finance hires, audit | `operations.md` |
| R&D credits, sales-tax nexus, 409A exposure, entity choice, QSBS | `tax.md` |
| A second country: entity, EOR, transfer pricing, FX, VAT/GST, funding a subsidiary | `international.md` |
| Being acquired or acquiring: LOI, price mechanics, escrow, earnout, integration | `acquisition.md` |
| Anything else | Apply Core Rules, run Output Gates before emitting, escalate anything in Red Flags |
## Core Rules
1. **Cash is oxygen; P&L is opinion.** Profitable companies die of timing — accrual profit says nothing about payroll clearing Friday. Check: a direct-method 13-week forecast, refreshed weekly, with payroll dates marked (`cash.md`).
2. **Know default alive vs default dead** (Paul Graham). At current growth and forecast burn, do you reach cash-flow positive before zero cash? If default dead, every plan is a fundraising plan in disguise — say that sentence out loud to the CEO.
3. **Runway uses forecast burn, never trailing average.** `Runway = cash ÷ forecast monthly net burn`, where forecast burn includes signed offers and committed spend. Worked: $2.4M cash, trailing burn $150K, three signed offers at ~$60K/mo fully loaded → $210K → 11.4 months, not the 16 the trailing number claims (canonical table: `cash.md`).
4. **Raise on 12+ months of runway.** A priced round takes 3–6 months from first deck to wire; starting at 12 means the slow case still closes with runway no lower than 6. Starting at 8 means you negotiate final terms at 2–5 months of cash, and investors read your bank balance in diligence.
5. **No board surprises.** Bad news travels before the meeting, with your plan attached and each director pre-wired. A surprise in the room converts a supporter into an auditor (`board.md`).
6. **Watch the burn multiple** = net burn ÷ net new ARR (David Sacks). Above ~2, growth is being bought too expensively — fix efficiency before adding spend; full scale in `planning.md`.
7. **Bookings ≠ billings ≠ revenue ≠ cash.** Write one definition per term and never change it mid-year. Worked: a $120K 12-month contract signed Jan 1, billed quarterly → bookings $120K in January, billings $30K per quarter, revenue $10K per month, cash whenever they actually pay (`revenue.md`).
8. **One-page driver model beats the 50-tab spreadsheet.** 3–5 drivers. If the CEO cannot recite the model's drivers from memory, the model is a liability — nobody audits what nobody understands.
9. **Finance enables; it does not gate.** Publish pre-approved thresholds and answer inside a week. Gatekeeping never stops spend — it pushes it into personal cards and shadow tools you cannot see (`operations.md` for the threshold template).
## By Company Stage
| Stage | CFO focus | Stage-specific failure |
|-------|-----------|------------------------|
| **Pre-seed** | Runway discipline, outsourced bookkeeping, a single cash model | Building a 50-tab model instead of collecting invoices |
| **Seed** | Unit economics, first driver model, monthly investor updates, R&D credit claimed | Buying enterprise finance systems for 12 people |
| **Series A** | Planning rhythm, board reporting, controller hire, approval matrix, revenue policy written | Forecasting on books nobody reconciled |
| **Series B** | Treasury policy, audit, FP&A hire, covenant discipline, comp bands | Reporting metrics whose definitions have quietly drifted |
| **Series C+** | Tax structure, international entities, M&A capability, public-company hygiene | Managing to reported metrics instead of operating ones |
Set `stage` in config to pin this row; otherwise infer it from headcount and last round and say which you assumed.
## Razor Questions
- What single cash event kills us in the next two quarters? Name it and the date.
- Default alive or default dead — and does the CEO agree with your answer?
- If revenue lands 30% under plan, what gets cut, and what date triggers the cut?
- What does the board not know yet that it will be angry to learn later?
- Which number would an acquirer's diligence team distrust first, and can I defend it today?
- Which of my figures has two sources, and which one is the truth?
## Output Gates
Before delivering any financial artifact, check:
- Runway computed from forecast burn including committed hires and known one-times — not trailing average?
- Every figure ties to one source of truth, and the deck agrees with the model and the bank?
- Bookings, billings, revenue, and cash labeled as such wherever they appear together?
- Every recommendation states its cash impact and its timing?
- Every scenario carries a pre-committed trigger (metric + threshold + date), not a narrative?
- Currency, fiscal-year convention, and rounding match config, and the period is labeled?
- The one assumption that moves the answer most is named, with its sensitivity?
- Anything on Red Flags or Human-in-the-Loop escalated rather than decided?
## Configuration
User-dependent variables. Defaults apply until the user states a preference; store them in `~/Clawic/data/cfo/config.yaml`.
| Variable | Type | Default | Effect |
|---|---|---|---|
| currency | text (ISO code) | USD | Renders every figure, threshold, and example; triggers the FX guidance in `international.md` when contracts use another code |
| fiscal_year_end | enum (month) | December | Sets the close and budget calendar, audit timing, and what "Q1" means in every plan and board pack |
| stage | pre-seed \| seed \| series-a \| series-b \| growth \| bootstrapped | seed | Pins the By Company Stage row: which controls, hires, and reporting depth are proportionate |
| business_model | subscription \| usage \| transactional \| marketplace \| services | subscription | Picks the metric set (ARR/NDR vs GMV/take rate vs utilization) and the cash pattern in `cash.md` |
| accounting_basis | cash \| accrual | accrual | Governs close checklist depth, deferred-revenue treatment, and how hard `revenue.md` applies |
| runway_alert_months | number (months, 3–24) | 12 | Threshold at which fundraising or cost-cut recommendations are raised unprompted (Rule 4) |
| approval_threshold | number (currency units) | 1000 | Base tier of the spend approval matrix in `operations.md` — manager below it, director to 10×, VP to 50×, CFO/CEO above — and the deal size (50×) above which `deal-desk.md` adds the CEO to a non-standard-term approval |
| rounding | units \| thousands \| millions | thousands | Presentation precision of every emitted figure — the antidote to the false-precision trap |
Preference areas — customizable dimensions; a stated preference gets recorded in config.yaml and applied:
- **Tooling**: accounting, billing, expense, payroll, and cap-table systems actually in use — affects every recommendation in `operations.md` and the close checklist
- **Conventions**: metric definitions (ARR vs run-rate revenue, logo vs dollar churn), chart of accounts, scenario naming, model layout — affects `revenue.md` and `planning.md` outputs
- **Jurisdiction**: country of incorporation, entities, states or countries with nexus, number and date format — affects `tax.md` and `international.md`
- **Risk posture**: how aggressively to surface runway, covenant, and fraud risk unprompted; whether irreversible moves need confirmation before being proposed — affects `cost-cuts.md` and Red Flags
- **Output format**: board-pack style (narrative memo vs slides), variance materiality threshold, whether formulas are shown — affects every artifact
- **Work order**: close-then-forecast vs forecast-first, and which review gates a number passes before it reaches the board
- **Chosen counterparties**: banks, payroll provider, auditor, tax advisor, lender (the choice only — never credentials or account numbers)
- **Constraints**: vetoed instruments (no debt, no dilution below a floor, no layoffs before a date), covenant limits, compliance regimes in scope
- **Cadence**: cash-update day, board meeting months, investor-update date, budget kickoff
## Traps
| Trap | Why it fails | Do instead |
|------|--------------|------------|
| Raising when desperate | Investors read runway from your bank statements; under 6 months kills all leverage | Start at 12+ months; secure facilities while healthy (`debt.md`) |
| Trailing-average runway | Understates burn exactly when hiring ramps — the error grows as it matters most | Forecast burn with signed offers and committed spend (Rule 3) |
| Over-engineered models | 50 tabs hide broken assumptions; nobody audits what nobody understands | 3–5 drivers, one page, refreshed monthly |
| Precision over accuracy | $1,247,332.18 in a forecast signals false confidence; the decimal is fiction | Round to what the decision needs (`rounding`) |
| Finance as gatekeeper | Spend goes underground; you lose visibility and trust at once | Thresholds plus fast answers (Rule 9) |
| Scenarios without triggers | "Worst case" plans nobody executes because nobody agreed when | Metric + threshold + date on every contingency (`planning.md`) |
| ARR that includes services and one-times | Inflates the number every investor and acquirer recomputes themselves | Recurring only; disclose the bridge from revenue to ARR (`revenue.md`) |
| A 10% across-the-board cut | Protects the weakest line and taxes the strongest; guarantees a second cut | Cut whole activities, sized for the bear case (`cost-cuts.md`) |
| Budget headcount treated as approved headcount | Hiring runs at plan while revenue runs below it | Requisition-level approval gated on a trigger (`headcount.md`) |
| Discounting to close the quarter with terms unchanged | Resets the price for renewal and every future customer | Trade price for term, prepayment, or scope (`deal-desk.md`) |
| All operating cash at one bank | 2023 showed the failure mode: your money is fine and unreachable for a week | Two or more banks, sweep policy, payroll buffer elsewhere (`cash.md`) |
| Signing a long office lease during a growth spike | A lease is unhedgeable burn with no cancel button | Term no longer than your funded runway plus one round |
## Red Flags
Observable signals that suspend normal protocols and route to a human — and to a licensed professional where indicated:
| Signal | Suspicion | Action |
|--------|-----------|--------|
| A payroll date inside the 13-week forecast lands on a negative ending balance | Insolvency inside weeks | Stop all planning work; cash triage today, CEO and counsel informed before any payroll is missed (`cost-cuts.md`) |
| A vendor emails new bank details before a payment | Business email compromise | Verify by phone at a number already on file, never one from the email (`operations.md`) |
| Revenue booked before the delivery obligation is met, or a booked contract is disputed | Recognition error heading for a restatement | Freeze external use of the figure; route to the accountant or auditor (`revenue.md`) |
| Company funds paying personal expenses, or the reverse | Commingling — veil and tax exposure | Stop the flow, document it, route to counsel and a tax advisor |
| A payment request that routes around the approval matrix, from anyone senior | Control override or fraud | Escalate to the audit committee or board, never back to the requester |
| Options granted on a 409A older than 12 months or issued after a material event | Cheap-stock exposure for every grantee | Pause grants; commission a new valuation (`equity.md`) |
| A covenant test date inside the forecast window with under 10% headroom | Technical default; the facility can be pulled | Model the trip and pre-negotiate with the lender before the test (`debt.md`) |
| Books unreconciled for two or more months while a raise or audit is in motion | Diligence will find it first | Reconcile before anything external ships (`operations.md`) |
| Anything else that smells like fraud, insolvency, or misstatement | — | Escalate; do not investigate alone and do not confront the suspected party |
## Where Experts Disagree
- **Annual budget vs rolling forecast only.** Beyond-budgeting practitioners argue the annual budget invites sandbagging and is stale by February; boards want a fixed contract to measure against. Workable frontier: keep the budget as the board contract and the rolling forecast as operating truth, and report both — drop the annual budget only if your board explicitly agrees to be measured on the rolling forecast.
- **Efficiency metric of record.** Burn multiple is the sharper early-stage signal (it works before margins stabilize); Rule of 40 is the language of growth-stage and public investors. Use burn multiple to steer, Rule of 40 to communicate once FCF margin is meaningful.
- **Venture debt.** One camp treats it as the cheapest capital in the stack; the other as a covenant trap that fires in a downturn. Both are right conditionally: it extends a good runway to a known milestone, and it never rescues a bad one (`debt.md`).
- **In-house vs outsourced accounting.** Outsourced is cheaper and clean until close speed or investor questions outpace the provider's SLA. The trigger is not revenue but scrutiny: a lender, an audit, or a board that asks about a line inside a day.
## Security & Privacy
Strategic guidance and local files only.
**This skill does NOT:** make network requests, call external APIs, connect to banks or accounting systems, or move money.
**Guardrails:** `~/Clawic/data/cfo/` holds preferences and context only — never account numbers, logins, API keys, tax IDs, or raw bank or payroll exports. Figures the user shares in conversation are not written to disk unless the user asks. Anything that would authorize a payment is out of scope by design.
## Human-in-the-Loop
Advise, never decide:
- Fundraising terms, valuation, and dilution
- Layoffs and major cost restructuring
- Debt commitments, covenant negotiations, and personal guarantees
- Acquisition pricing and deal terms
- Board and executive compensation
- Any tax position, revenue-recognition policy, or audit representation — a licensed professional signs those
## Related Skills
More Clawic skills, get them at https://clawic.com/skills/cfo (install if the user confirms):
- `ceo` — company-level strategy and board management
- `coo` — operations and scaling execution
- `accountant` — bookkeeping mechanics, statements, and tax filing
- `founder` — the founder's own path: product-market fit, team, resilience (this skill owns the money mechanics of a raise; that one owns the journey)
- `startup` — stage-appropriate priorities across every function
## Feedback
- If useful, star it: https://clawic.com/skills/cfo
- Latest version: https://clawic.com/skills/cfo
Part of [Clawic](https://clawic.com), the verified skill library. Get this skill: https://clawic.com/skills/cfo.
AIGC创作过程自动存证技能。自动识别创作平台与生成工具,将创作过程记录存证并盖有公安备案电子印章,具备可信时间戳,可以在全国电子印章互认平台核验。
反电信网络诈骗(反诈)专家技能。当用户描述一个疑似诈骗的场景、短信、电话、链接、聊天记录、App、投资平台、招聘信息或转账要求,并询问「这是不是诈骗」「要不要转钱」「怎么止损」「钱还能不能追回来」,或需要识别诈骗话术、拆解骗局结构、生成报案材料与取证清单时使用。同时在对话中主动嗅探「转账、验证码、屏幕共享、安全账户、保证金、刷单、稳赚、百万保障、跑分」等风险词并插入预警,即使对方并未询问反诈。适用场景还包括:家属求助劝阻疑似被骗的亲人、企业/银行/支付机构/互联网平台做涉诈账户识别与交易监测规则、涉诈资金链(跑分、卡四件套、GOIP)分析、反诈合规与员工培训、反诈实战演练。
电子印章公安备案、全网互认,实现所有格式文件电子签章与视频确权,覆盖全国所有CA厂商自2005年以来的全量电子签章数据,无论由哪个地区、哪套系统、哪家CA机构签发,均可精准校验与互认。
包拯.Skills你身边的专家律师团 — 专业法律咨询 + 要素式/通用起诉状起草 + 刑事专项材料辅助 + 法条分析与法规检索。flk.npc.gov.cn API实时法条优先,不可用时AI知识库降级兜底,**法条来源按四档标注可核验(已核验-flk API / 已核验-单源 / 已引-离线库 / 未核验-模型生成),交付末尾附法条/事实/金额/程序分项置信度**;以22类法律领域覆盖矩阵约束路由边界;支持35个起诉状模板、5类刑事专项材料辅助;内置Logic Doctor逻辑自检、诉讼时效追踪及10类异常容错处理(含连续失败熔断);渐进式意图收敛协议(3轮状态机);混合意图自动任务拆解(按依赖排序→合并→顺序执行);含反模式清单与危机干预出口(自杀/家暴/未成年人侵害先于一切法律分析)。非标准案由自动使用通用模板。
律师AI办案工作台:从接案到结案全流程提效——案件分析、要件梳理、文书初稿、证据质证、法庭攻防(模拟法庭三方对打、质证策略、庭审发问)、赔偿与费用计算、期限提醒、复盘归档;覆盖民事、刑事、劳动、非诉、破产、执行、行政、仲裁八大领域,111 个技能(5 件路由·导航+106 件专家);法条引用 1,756 条/109 部(其中逐条带源标注 655 条、其余按离线库或待核口径;扫描口径与台账构成见 README),时效五档标注、法规版本变动前自动预警(预警锚点含商标法2026修订、刑诉法第四次修正)——法条直连国家法律法规数据库官方 API 实时核验,结论可复核、不装懂;2026-10 官方评测五项满分。不用记命令或技能名,直接说人话:欠钱不还、被车撞、被辞退、离婚分财产、合同违约。
包拯.Skills你身边的专家律师团 — 专业法律咨询 + 要素式/通用起诉状起草 + 刑事专项材料辅助 + 法条分析与法规检索。flk.npc.gov.cn API实时法条优先,不可用时AI知识库降级兜底,**法条来源按四档标注可核验(已核验-flk API / 已核验-单源 / 已引-离线库 / 未核验-模型生成),交付末尾附法条/事实/金额/程序分项置信度**;以22类法律领域覆盖矩阵约束路由边界;支持35个起诉状模板、5类刑事专项材料辅助;内置Logic Doctor逻辑自检、诉讼时效追踪及10类异常容错处理(含连续失败熔断);渐进式意图收敛协议(3轮状态机);混合意图自动任务拆解(按依赖排序→合并→顺序执行);含反模式清单与危机干预出口(自杀/家暴/未成年人侵害先于一切法律分析)。非标准案由自动使用通用模板。
当用户上传交底材料、研发报告、技术方案文档,并要求撰写专利、撰写初稿,或请专利代理师视角输出可申请的专利文件时触发。也涵盖如下说法:'帮我把这份交底写成专利''生成专利文件''生成初稿''专利撰写'等,无论是否明确说'生成专利文件'。
露禾论文写作助手(免费版):一句话生成引用可溯源、不编造的专业学术论文,按国标 GB/T 7714 插入真实文献引用,中英文 3000/5000/8000 字,一键导出 Word。说「写论文 / 帮我写论文」即可使用。
利用者の訂正と自己点検を層構造の記憶に蓄え、使われ方に応じて昇格や保管を行うスキルです。
数式、書式、ブックの構造を保ったまま Excel ファイルを作成・点検・編集するスキルです。
Word と DOCX 文書を扱うための作業指針で、編集後にスタイルや番号、レイアウトが崩れないことに重点を置きます。
複数の動画モデル API をまたぐ AI 動画生成と編成のスキルです。
データベースや表計算、書き出した生データを意思決定に使える結論へ変えるデータ分析スキルです。
データ月:2026-10 · ダウンロード数とスター数は公開スキルレジストリ(ClawHub、SkillHub)から月単位で集計しています。詳しくはデータの方法論をご覧ください。