PDN SKILL · v0.4.7
A Skill for understanding how a business makes money
A free conversational tool for clarifying a business situation, separating evidence from hypotheses and unknowns, and choosing what to check next. Read the sample first; install the diagnosis Skill only if you want to try it yourself.
What is PDN Skill?
PDN Skill is a free add-on for Claude regular chat. It guides Claude through questions about your business, then organizes what is known, what may be true, and what still needs checking.
Claude can answer many questions, but conversations about running a company can end with general advice. With PDN Skill, it first asks about your company and separates what is known, what may be true, and what is still unknown.
It avoids jargon where possible. By answering questions, you get a document to take away and use as a starting point when consulting an accountant, bank, or other professional.
What happens when I use it?
- Describe a business concern in your own words, such as “new orders are declining.”
- The AI asks one question at a time. Answer what you can; say “I don't know” when you don't. That uncertainty is included in the summary.
- After several questions and answers, the AI separates what is known, possible causes, and what still needs checking.
- The result is a document you can keep.
- If you want, it can also draft a business plan or a monthly cash outlook from that summary.
Who is it for?
- Business owners and sole proprietors who want to organize their company's numbers and structure.
- People who want to describe their business situation before consulting a professional.
No accounting or management expertise is required.
Supported AI
- Claude: available across plans including Free, subject to each plan's limits.
- ChatGPT: individual Free and Plus plans do not include the Skills feature, so PDN Skill cannot be installed there. Eligible business and organization plans may support it depending on administrator settings.
This page gives setup steps only for supported environments.
Cost
PDN Skill is free to use within its license. AI service charges follow your provider's plan.
Limitations
PDN Skill is not a business planning or accounting professional. Figures are hypotheses at that point in time. Consult an accountant or other professional before important decisions.
Downloads
No account or email address is required. The diagnosis Skill is the one to start with.
- pdn-profit-diagnosis v0.4.7 — SHA-256
3d2981eeccd474d65e4f9dc1d98a0774ff5346356081fe610e8c7336eaa2fa25 - pdn-profit-diagnosis (Muse) v0.4.7 — SHA-256
27b7e5d9853f8a4903e93a4020a8824288f42ae750cffe8ca5fe2d5dc2606165 - pdn-business-planning v0.4.7 — SHA-256
a37b721b4b0202cff305c9af099172eb675e71f040b01127fc3bcd7ef2c4c453 - pdn-business-planning (Muse) v0.4.7 — SHA-256
b5c1f2ceddbc402314b89cce3485873c3adb49d766c184d71d082b9884830105 - pdn-accounting-draft v0.4.7 — SHA-256
9b0d8a71c79f88a5c8ec06d63bc5bbc2ac5afcadae1f1f73c9abd0a3a58a8656 - pdn-accounting-draft (Muse) v0.4.7 — SHA-256
21e77ea9e3ad2bf5d02bf97d338b2842d1c135eb468fc4aa800a92f2fb30520d
Supported environments and review date
Reviewed November 21, 2026. Feature availability and administrator settings can change.
- Claude regular chat: Available on plans including Free, Pro, Max, Ultra, Team and Enterprise. Enable code execution and file creation.
- Claude Code: Also supported in the developer command-line environment; see the developer setup note below.
- Muse: Supported by installing the Muse ZIP through a chat with your Muse.
- ChatGPT: Organization plans may offer Skills depending on administrator settings.
- Codex: Developer-specific setup.
Use in regular Claude chat
- Open Claude Settings > Capabilities and enable code execution and file creation.
- Open Customize > Skills, then choose + > Create skill > Upload a skill.
- Select the downloaded ZIP without extracting it, upload it, and turn the Skill on.
- Confirm the Skill is enabled in the list. In a new chat, ask for a task that names it; the Skill should appear in the response's Skills indicator.
If setup does not work, check that code execution and file creation are enabled, that you selected the ZIP without extracting it, and that you tested in a new chat.
Start with pdn-profit-diagnosis. Business planning and accounting draft are optional and can be added separately.
For Muse, download a Muse ZIP, attach it in a chat with your Muse, and say, “Please install this skill.” Muse will register it for use in regular conversations.
ChatGPT Free and Plus do not include Skills, so PDN Skill cannot be installed there. Ask your administrator about availability in an organization workspace.
For developers: install Skill files manually
For Claude Code and Codex, place the Skill files in these folders. In Claude Code, place the Skill folder at .claude/skills/<skill-name>/SKILL.md. In Codex, place it at .agents/skills/<skill-name>/SKILL.md.
Actual output from a regular Claude chat
This is the output from running a fictional case — a residential HVAC service company in suburban Houston, Texas — in a regular Claude chat on November 21, 2026. All figures are fictional test data. The run used v0.4.7, matching the version distributed on this page.
How to read this sample
- The free diagnosis is a sample of the approach. It is a hypothesis-stage summary: it separates confirmed facts, guesses, and unknowns, and it may not name the underlying mechanism.
- The business plan is a draft. In this run, the decision between hiring a fourth technician and subcontracting installations is deliberately left open until three checks are done (Gate A), and each milestone states what "done" means.
- The accounting forecast runs on placeholder cost assumptions, each marked as an assumption on its Assumptions sheet. Owner pay, loans, and truck payments are not included, so the profit shown is higher than the real situation — the forecast itself says so. Replace the placeholders with your own figures before relying on it.
- Outputs vary by AI host and conversation. The same case is not guaranteed to produce the same result.
Shareable takeaway
# HVAC Business Diagnosis — Takeaway Nov 21, 2026 · @Akinobu Fukuzaki [PDN Diagnosis] This is a working summary, dated 21 November 2026, of a diagnosis of a residential HVAC service company in suburban Houston. It is at the hypothesis stage: the facts below are the owner's own figures, but the explanations are not yet verified and the checks at the end are what will confirm or refute them. ## The business issue Installations, the company's highest-value work, have fallen about 29% in three years while summers stay full and cash feels tighter. Several causes may be acting at once, and which one dominates decides whether the fix is mainly about sales, pricing or cash timing. ## Confirmed facts All figures are the owner's, for last year unless noted. Shares and changes marked "calculated" are simple arithmetic on those figures.
| Revenue stream | Last year | Share (calculated) | Three years ago |
|---|---|---|---|
| Repairs | about $584,000 | 46% | not known |
| New installations | about $400,000 | 31% | about $560,000 |
| Maintenance agreements | about $296,000 | 23% | not known |
| **Total** | **about $1.28 million** | not known |
- **Installations:** down about $160,000, or about 29%, over three years (calculated). - **Agreements:** about 1,080 active at about $275 a year. Cancellations run about 3% a year and the count is flat, which implies roughly 30 new sign-ups a year (calculated). - **Bids:** the owner writes all estimates and is losing more of them to national chains advertising low prices. - **Owner's time:** age 58, about 55 hours a week, covering estimates and some field work. - **Team:** 3 field technicians at $32–38 an hour plus summer overtime; 1 part-time office person (20 hours a week) for billing and scheduling. - **Fixed costs:** shop and warehouse rent of $4,200 a month; 4 trucks. - **Seasons:** peaks in June–August (cooling) and December–January (heating); slow in March–April and October. - **Costs:** refrigerant, parts and labor prices are rising. - **Selling agreements:** whether a technician offers one on a repair visit is up to each technician. - **Never measured:** agreement attach rate, add-on sales to agreement customers, profit per agreement, revenue per technician, and lost quotes by customer type. ## Still a guess None of these is confirmed. Each is a possible explanation that fits the facts so far. 1. **Busy is not the same as profitable.** As installations shrink, more of the year's work is repairs, which usually bring in less per crew-hour. Phones stay full while each hour earns less. 2. **The agreement base may not be producing replacement jobs.** About 1,080 member homes should be the best source of new systems. If members are also getting quotes from chains when their system fails, the company loses the most valuable moment in a relationship it built. 3. **Some losses may be about speed, not price.** Every estimate depends on the owner, whose busiest field weeks are also the busiest quoting weeks. A chain that quotes the same day can win before a competitive quote goes out. 4. **The agreement base isn't growing.** Few cancellations suggest customers value the plan. Few new sign-ups, left to each technician, point to how agreements are offered. 5. **Cash may be tight for up to four reasons at once:** 1. less installation revenue 2. repair rates and the agreement price may have fallen behind rising costs 3. rent, trucks and technicians continue through three slow periods 4. when agreement money comes in compared with when bills are due There is also a longer-term point: installation sales run entirely through one person working about 55 hours a week. That limits growth and is a risk if the owner ever wants to step back. ## Important unknowns
| Question to verify | Evidence that would settle it | Which guess it tests |
|---|---|---|
| Are lost installation quotes mostly agreement members or new leads? | Last 12 months of estimates, each marked member / new lead and won / lost | 2 |
| Are quotes lost because they go out too slowly? | Days from first call to quote sent, on the same estimates | 3 |
| Have repair rates and the $275 agreement price kept up with costs? | Invoices for 2–3 common repairs and an agreement renewal from about three years ago vs. today, set against supplier invoices for the same parts | 5.2 |
| Do agreement customers pay yearly up front or monthly, and in which months do renewals fall? | Billing records, month by month, from the office person | 5.4 |
| Did repairs grow to fill the installation gap? | Repair and agreement totals from three years ago | 1 |
| What does one agreement visit cost to deliver? | Visits per agreement, typical time per visit including driving, and technician pay rate | 5.2 |
| How often do technicians offer an agreement on a repair visit? | A count of offers and sign-ups per technician over a season | 4 |
## Checks, in order of value The first two decide whether the main problem is sales or margin; the third settles cash timing. - [ ] **1. Estimate look-back.** For the last 12 months of quotes, mark each one member or new lead, won or lost, and days from call to quote. Even 30–40 quotes would show the pattern. - [ ] **2. Invoices then and now.** Compare 2–3 common repairs and an agreement renewal from about three years ago with today, and do the same for supplier costs on the same parts. - [ ] **3. Billing records with the office person.** Find out how agreements are paid (yearly or monthly) and how many renew each month. - [ ] **4. Technician notes, going forward.** At each tune-up, record any new system a member had installed by someone else. If most lost quotes are new leads, the fix is about competing for strangers. If many are members, the fix is inside the agreement program: spotting ageing systems early and quoting before a chain is asked. ## Context: typical maintenance-agreement pricing in the US These are general national ranges from public web sources, mostly cost guides and HVAC software or marketing firms, not an independent survey. Local labor rates shift them. They say nothing about whether $275 covers this company's own visit costs; check 2 does that.
| Plan type | Typical range | Usually includes | Source |
|---|---|---|---|
| Basic, one system | $150–$300 a year (most priced $175–$250) | Two tune-ups a year, spring cooling and fall heating | [SmartService](https://www.smartservice.com/blog/how-much-hvac-service-agreements) |
| Comprehensive, one system | $300–$500 a year | Two visits plus priority service, repair discounts, small extras | [SmartService](https://www.smartservice.com/blog/how-much-hvac-service-agreements) |
| Basic (2026 vendor figures) | $159–$219 a year | 1 system, 2 tune-ups, 10% off repairs | [Plyrium](https://www.plyrium.com/academy/hvac/hvac-maintenance-agreements-pricing-2026) |
| Standard (2026 vendor figures) | $219–$329 a year | 1 system, 2 tune-ups, 20% off repairs, priority | [Plyrium](https://www.plyrium.com/academy/hvac/hvac-maintenance-agreements-pricing-2026) |
| Each additional system | roughly $75–$250 a year | Depends on whether each unit gets a full tune-up | [LatestCost](https://latestcost.com/hvac-maintenance-contract-cost-u-s-homes/) |
- **Structure:** tiered good-better-best plans are common. Most are billed yearly; many also offer monthly payment, for example $199–$329 a year or $19–$29 a month ([Pipeline](https://pipelineon.com/blog/hvac-maintenance-agreement/)). Some add a 12-month minimum and an early-cancellation fee. - **Customer expectations:** homeowners expect at least two visits a year ([ACHR News](https://www.achrnews.com/articles/153922-residential-hvac-replacement-market-slowed-in-2025)). - **Price increases:** one long-running contractor charges $250 a year per system and raises it about 5–10% every other year ([ACHR News](https://www.achrnews.com/articles/153922-homeowners-contractors-mutually-support-maintenance-agreements)). - **Follow-on work:** software vendors commonly estimate about $2 of extra repair and replacement work per $1 of agreement value; this is a vendor estimate, not independent data ([Service Magazine](https://www.servicemag.org/guides/maintenance-agreements-recurring-revenue)). ## How this diagnosis thinks It maps how the business makes money first: what is sold, to whom, and which income repeats. It then traces causes from that structure, weighing evidence over impressions. Facts, guesses and unknowns are kept separate, so a guess is never acted on as if it were proven. ## When to revisit - When the results of check 1 or check 2 are in, even roughly. - Before a decision you need to explain to others, such as a price change, a hire or a new way of selling agreements. - When something you tried doesn't work and you want to know why. - Ahead of the next summer peak, or if cash tightens again in a slow month. PDN Skill v0.4.7
Download this output as Markdown
Read the business plan output
Business plan
# 3-Year Business Plan — Draft Nov 21, 2026 · @Akinobu Fukuzaki [PDN-Planning Draft] **Draft for the owner's review, as of 21 November 2026.** Assumed audience: the owner and anyone he chooses to share it with (an advisor, a lender, a key employee). It builds on the diagnosis summary of the same date and adds no figures beyond the owner's own. ## Summary - **Purpose:** recover installation work, turn the 1,080-home agreement base into a growing source of replacements, and steady cash through the slow months, over 2027–2029. - **Main figures (owner's):** revenue about $1.28M last year; installations about $400k, down from about $560k three years ago; about 1,080 agreements at about $275 a year. - **Key decision:** add capacity by hiring a fourth technician, or by subcontracting installation work. The plan recommends deciding **after** the first checks show which constraint is actually binding, and before the June–August 2027 peak. - **Biggest uncertainty:** whether lost installations are mainly lost members or lost new leads, and whether they are lost on price or on quoting speed. Most of the plan's direction turns on that. **Legend:** *Fact* = the owner's figure or statement. *Assumption* = a working belief to verify, listed in the register. *Unknown* = not yet known; shown as **[To confirm: …]**. *Proposed* = a suggestion in this draft, not a decision. ## Starting point A 14-year-old residential HVAC company in suburban Houston that is busy but has been losing its highest-value work.
| Item | Today (fact) | Status |
|---|---|---|
| Revenue, last year | about $1.28M: repairs \~$584k, installations \~$400k, agreements \~$296k | Fact |
| Installations, three years ago | about $560k | Fact |
| Agreements | about 1,080 active at \~$275/year; \~3% cancel a year; count flat | Fact |
| People | Owner (58, \~55 h/week, all estimates plus some field work); 3 technicians at $32–38/h plus summer overtime; office person 20 h/week | Fact |
| Fixed costs | Rent $4,200/month; 4 trucks | Fact |
| Seasons | Peaks Jun–Aug and Dec–Jan; slow Mar–Apr and Oct | Fact |
| Competition | Losing more bids to national chains advertising low prices | Fact |
| Costs | Refrigerant, parts and labor rising | Fact |
| Profit by job type, revenue per technician, cash by month | **[To confirm: from bookkeeping and billing records]** | Unknown |
**Causal picture carried from the diagnosis (still hypotheses):** the agreement base keeps crews busy with tune-ups and repairs but may not be producing replacement jobs. Some quotes may be lost on speed because every estimate runs through the owner. Prices may have fallen behind costs, and fixed costs run through three slow periods. ## Objectives Each objective is tied to the problem it addresses and the reason it should help. Targets are left open until the Year 1 checks set a baseline.
| # | Objective | Problem it addresses | Why it should work (hypothesis) |
|---|---|---|---|
| O1 | Win back installation work, starting with agreement members | Installations down \~29% in three years | Members already trust the company; reaching them before a chain does should lift capture |
| O2 | Grow the agreement base, not just hold it | Count flat; \~30 sign-ups a year (calculated) | A consistent offer on every repair visit should raise sign-ups while cancellations stay low |
| O3 | Make prices cover today's costs | Possible price lag behind rising costs | Regular, recorded price reviews stop the margin eroding a little each year |
| O4 | Even out cash across the year | Fixed costs run through three slow periods | Agreement billing and visit scheduling can move money and work into the slow months |
| O5 | Reduce dependence on the owner for estimates | All quotes run through one person at \~55 h/week | A second person able to quote should cut response time and lower key-person risk |
## Roadmap Year 1 measures and fixes, Year 2 builds, Year 3 makes it last. Each year opens only when the gate before it is passed. ### Year 1 (2027): measure, then fix the cheapest levers
| Milestone | Done when | Objective | Proposed timing |
|---|---|---|---|
| Estimate look-back complete | Last 12 months of quotes marked member/new lead, won/lost, days to quote | O1, O5 | Before the March–April slow period |
| Price check complete | 2–3 common repairs and an agreement renewal compared then vs. now, against supplier costs | O3 | Before the March–April slow period |
| Billing timing known | Agreement payment method and renewals by month listed with the office person | O4 | Before the March–April slow period |
| Agreement offer made standard | A written offer every technician uses on repair visits; offers and sign-ups counted | O2 | Spring 2027 |
| Capacity decision made | Gate A passed (below) and one option chosen | O1, O5 | Before June 2027 |
| Price review done | New repair and agreement prices set, or a documented decision to hold | O3 | **[To confirm: owner to set date after the price check]** |
**Gate A (into the capacity decision):** the three checks are complete. Then choose: if lost quotes cluster on speed or members, fix quoting first; if field hours in peak are the limit, add capacity (see the decision section). ### Year 2 (2028): build the replacement pipeline
| Milestone | Done when | Objective |
|---|---|---|
| Member replacement program running | Ageing member systems are flagged at tune-ups and quoted before failure | O1 |
| Second person able to quote | Someone other than the owner sends standard replacement quotes | O5 |
| Capacity option reviewed after one full peak | Year 1 results compared against the continue/adjust/stop criteria | O1 |
| Slow-month plan in place | Agreement visits and billing scheduled to fill March–April and October | O4 |
**Gate B (into Year 3):** member capture, quote speed and sign-ups have moved in the right direction against the Year 1 baselines. If not, revisit the causes before scaling. ### Year 3 (2029): make it last
| Milestone | Done when | Objective |
|---|---|---|
| Owner's role narrowed | Owner works mainly on estimating oversight and the business, not daily field work | O5 |
| Annual price review routine | Prices reviewed and recorded once a year | O3 |
| Succession options outlined | The owner has written down his intentions for the business's future | O5 |
## Pending decision: fourth technician or subcontracted installations **Proposed:** decide after Gate A, before June 2027. First establish which constraint is binding: field hours in peak, the owner's quoting time, or bids lost on price. Neither option helps if the real cause is lost bids. Subcontracting installations adds installation capacity, but installations are falling because bids are lost, not, as far as we know, because crews can't fit them in.
| Option A: hire a fourth technician | Option B: subcontract installation work | |
|---|---|---|
| **Pros** | Adds field hours in both peaks. Can take field work off the owner, freeing time to quote faster. Company controls quality and the customer relationship. Another person who can offer agreements on visits. | Cost moves with the work: paid only when there is an installation. No hiring needed, which avoids the local shortage of experienced technicians. Frees own technicians for repairs and agreement visits. Quick to start or stop. |
| **Cons** | Experienced technicians are hard to hire locally. Wages run year-round, including the three slow periods. Training and supervision would likely fall on the owner. | Less profit per installation, since the subcontractor takes a share **[To confirm: subcontractor quotes]**. Less control over quality, warranty work and the customer's experience. Subcontractors are likely busiest in the same peaks. Doesn't reduce the owner's estimating load. |
| **Main risks** | A long or failed search. A hire who doesn't meet the standard. Paying for idle hours in March–April and October, which adds to the cash strain. | Poor workmanship or callbacks under the company's name. Members switching to the subcontractor or a chain. No crew available in a heat wave. |
| **What must be true** | Someone suitable can be hired **[To confirm: owner's sense of the local hiring market and time to fill]**. There is enough paid work to keep them busy outside the peaks, for example agreement visits moved into slow months. Prices cover the added wage (price check). Peak field hours, not quoting, are the bottleneck. | A reliable, licensed subcontractor is available in the peaks. The remaining margin is acceptable **[To confirm: subcontractor rates]**. The company keeps the quote, the customer and the warranty. Installation volume recovers enough to need extra crews. |
| **Continue / adjust / stop** | Review after the first full peak: continue if peak backlog falls and the owner's quote turnaround improves. Adjust (move visits into slow months) if slow-month hours sit idle. Stop or don't backfill if the hire's work doesn't cover its cost. | Pilot on a few jobs first. Continue if callbacks and customer complaints stay at the company's normal level and margin is acceptable. Stop if quality slips or members go elsewhere. |
**Variants worth weighing:** hiring a less-experienced technician and training them, if the owner or a senior technician has the time. Subcontracting only peak-season overflow, not all installations. Moving some estimating to a second person (O5), which tackles quoting speed whichever option is chosen. ## Measures to track Leading measures move first and show whether the plan is working. Lagging measures confirm the result later. All targets are **proposed** and should be set once a baseline exists. ### Leading
| Measure | How it's calculated | Source | How often | Owner | Baseline | Target |
|---|---|---|---|---|---|---|
| Quote turnaround | Days from first call to quote sent | Estimate records | Monthly | Owner | Unknown (check 1) | Proposed: set after check 1 |
| Quote win rate, members vs new leads | Quotes won ÷ quotes sent, split by customer type | Estimate records | Monthly | Owner | Unknown (check 1) | Proposed: set after check 1 |
| Agreement offer rate | Repair visits with an agreement offered ÷ all repair visits | Technician job notes | Monthly | Each technician; reviewed by owner | Never measured | Proposed: set after 3 months of counting |
| Agreement sign-ups | New agreements per month | Billing records | Monthly | Office person | About 30 a year (calculated) | Proposed: set in Year 1 |
| Members' replacements lost | New systems found at tune-ups that the company didn't install | Technician job notes | Monthly | Technicians | Never measured | Proposed: lower each year |
### Lagging
| Measure | How it's calculated | Source | How often | Owner | Baseline | Target |
|---|---|---|---|---|---|---|
| Installation revenue | Total installation billing | Bookkeeping | Quarterly | Owner | About $400k last year | Proposed: set after Gate A |
| Active agreements | Count at period end | Billing records | Quarterly | Office person | About 1,080 | Proposed: growth set in Year 1 |
| Agreement cancellations | Cancelled ÷ active at start of year | Billing records | Yearly | Office person | About 3% | Proposed: hold at or below today |
| Revenue per technician | Revenue ÷ field technicians | Bookkeeping | Quarterly | Owner | Never measured | Proposed: set in Year 1 |
| Cash at the end of each slow period | Bank balance end of April and October | Bank statements | Twice a year | Owner | **[To confirm: last two years' balances]** | Proposed: set with an accountant |
## Assumptions to check
| # | Assumption | Basis | How much the plan depends on it | How it's verified | By when |
|---|---|---|---|---|---|
| A1 | A meaningful share of lost installation quotes are agreement members | Diagnosis hypothesis; not measured | High: shapes O1 | Check 1 (estimate look-back) | Before March–April 2027 |
| A2 | Some quotes are lost on speed, not only price | Owner does all estimates while also in the field | High: shapes O5 and the capacity decision | Days to quote in check 1 | Before March–April 2027 |
| A3 | Repair rates and the $275 price have fallen behind costs | Costs rising; no record of last price change | High: shapes O3 and whether a new hire pays | Check 2 (invoices then and now) | Before March–April 2027 |
| A4 | Agreement billing timing leaves slow months short of cash | Not yet known | Medium: shapes O4 | Check 3 (billing records) | Before March–April 2027 |
| A5 | A suitable fourth technician can be found locally | Owner reports experienced technicians are hard to hire | High for Option A | Owner's hiring search or local enquiries | Before June 2027 |
| A6 | A reliable subcontractor is available in peaks at an acceptable rate | Not yet known | High for Option B | Subcontractor quotes and references | Before June 2027 |
| A7 | Technicians will offer agreements consistently once it's standard | Not yet known | Medium: shapes O2 | Offer rate after 3 months | Summer 2027 |
| A8 | Someone other than the owner can learn to write standard quotes | Not yet known | Medium: shapes O5 | Trial with one person on simple replacements | Year 2 |
## Risks
| Risk | Early-warning sign | Likely effect | Response | Owner |
|---|---|---|---|---|
| Owner overload stalls the plan | Checks slip past the slow period; quote turnaround grows in peak | Year 1 decisions made without evidence | Do the checks in March–April; hand data gathering to the office person where possible | Owner |
| Capacity added before the cause is known | Hiring or subcontracting starts before Gate A | Higher costs with no installation recovery | Hold the decision until Gate A is passed | Owner |
| Price increase loses members | Cancellations rise above today's \~3% after a change | Smaller agreement base | Explain the change to members; review after one renewal cycle | Owner, office person |
| Hire not found, or doesn't work out | No suitable candidate by spring 2027 | Another peak without added capacity | Fall back to peak-only subcontracting or a trainee | Owner |
| Subcontractor quality slips | Callbacks or complaints on subcontracted jobs | Damage to reputation and member trust | Pilot small; keep the right to end the arrangement | Owner |
| Cash squeeze in a slow month | Bank balance falling through March–April or October | Difficulty covering payroll and rent | Use check 3 results to plan billing; discuss options with an accountant | Owner |
| Key-person risk | Owner unavailable for a period | Quoting and field work stop | Train a second person to quote (O5); write down core processes | Owner |
## First 90 days These steps fit around the December–January heating peak and set up Gate A before spring. - [ ] Start marking every new quote member/new lead, won/lost and days to quote (Owner) - [ ] Ask technicians to note at each visit any member system installed by someone else (Technicians) - [ ] Pull agreement payment method and renewals by month (Office person) - [ ] Complete the 12-month estimate look-back (Owner) - [ ] Compare 2–3 common repairs and an agreement renewal, then vs. now, against supplier costs (Owner) - [ ] Write a standard agreement offer and start counting offers on repair visits (Owner, technicians) - [ ] Get initial quotes from one or two possible subcontractors, and sound out the local hiring market, without committing (Owner) - [ ] Review results against Gate A and choose the capacity option (Owner) **Items needing outside review:** pricing changes, hiring costs and cash planning may benefit from an accountant's input; agreement terms and subcontract terms may need legal review. This draft prepares those discussions but does not replace them.
Read the accounting output
Accounting spreadsheet transcription
### Sheet: Assumptions
| Item | Value | Unit | Status | Note |
|---|---|---|---|---|
| Repairs revenue, last year | 584,000 | $ / year | Given | Your figure |
| Installation revenue, last year | 400,000 | $ / year | Given | Your figure |
| Active maintenance agreements | 1,080 | count | Given | Your figure |
| Agreement price | 275 | $ / year | Given | Your figure |
| Revenue change vs last year | 0 | % | Assumption | Flat. Change this to test a better or worse year |
| Peak-month weight | 1.5 | × a normal month | Assumption (illustrative) | How much busier a peak month is than a normal one. Not measured — replace from your monthly sales if you have them |
| Normal-month weight | 1 | × a normal month | Assumption | Reference level |
| Slow-month weight | 0.6 | × a normal month | Assumption (illustrative) | Not measured — replace from your monthly sales if you have them |
| Field technicians | 3 | people | Given | Your figure |
| Technician pay rate | 35 | $ / hour | Assumption | Midpoint of your $32–38 range |
| Regular hours per technician | 173 | hours / month | Assumption | 40 h × 52 weeks ÷ 12, paid every month including slow ones |
| Overtime hours per technician, peak months | 20 | hours / month | Assumption (illustrative) | You said there is summer overtime; the amount was not given. Applied to every Peak month |
| Overtime pay multiplier | 1.5 | × pay rate | Assumption | Time-and-a-half; confirm |
| Office person hours | 87 | hours / month | Derived | Your 20 h/week × 52 ÷ 12 |
| Office person pay rate | 20 | $ / hour | Assumption (illustrative) | Not given — replace |
| Payroll taxes and benefits | 0.1 | % of wages | Assumption (illustrative) | Not given — confirm with your accountant |
| Installation equipment and materials | 0.45 | % of installation revenue | Assumption (illustrative) | Not given — replace from supplier invoices |
| Repair parts and refrigerant | 0.25 | % of repair revenue | Assumption (illustrative) | Not given — replace from supplier invoices |
| Agreement visit materials | 0.05 | % of agreement revenue | Assumption (illustrative) | Filters, cleaning supplies etc. Not given — replace |
| Shop and warehouse rent | 4,200 | $ / month | Given | Your figure |
| Trucks | 4 | count | Given | Your figure |
| Running cost per truck | 800 | $ / month | Assumption (illustrative) | Fuel, insurance, maintenance, any payments. Not given — replace |
| Other overhead | 3,000 | $ / month | Assumption (illustrative) | Business insurance, phones, software, marketing, accounting. Not given — replace |
| Opening cash balance, 1 Dec 2026 | 0 | $ | To confirm | Enter your bank balance. Left at 0, closing cash shows only the change over the year |
| Monthly pattern — season of each month and share of annual agreement money collected | ||||
| Month | Season | Agreement cash collected (% of year) | Status | |
| Dec 2026 | Peak | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Jan 2027 | Peak | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Feb 2027 | Normal | 0.08 | Season: assumption. Cash share: to confirm (check 3) | |
| Mar 2027 | Slow | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Apr 2027 | Slow | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| May 2027 | Normal | 0.08 | Season: assumption. Cash share: to confirm (check 3) | |
| Jun 2027 | Peak | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Jul 2027 | Peak | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Aug 2027 | Peak | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Sep 2027 | Normal | 0.08 | Season: assumption. Cash share: to confirm (check 3) | |
| Oct 2027 | Slow | 0.08 | Season: given. Cash share: to confirm (check 3) | |
| Nov 2027 | Normal | 0.08 | Season: assumption. Cash share: to confirm (check 3) | |
| Total (must be 100%) | 1 |
### Sheet: Forecast
| 12-month forecast, Dec 2026 – Nov 2027 (USD) — draft for review; every figure is driven by the Assumptions sheet | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Line | Dec 2026 | Jan 2027 | Feb 2027 | Mar 2027 | Apr 2027 | May 2027 | Jun 2027 | Jul 2027 | Aug 2027 | Sep 2027 | Oct 2027 | Nov 2027 | Total |
| Season | Peak | Peak | Normal | Slow | Slow | Normal | Peak | Peak | Peak | Normal | Slow | Normal | |
| Seasonal weight | 1.5 | 1.5 | 1 | 0.6 | 0.6 | 1 | 1.5 | 1.5 | 1.5 | 1 | 0.6 | 1 | |
| Share of year's repair and installation work | 0.11 | 0.11 | 0.08 | 0.05 | 0.05 | 0.08 | 0.11 | 0.11 | 0.11 | 0.08 | 0.05 | 0.08 | 1 |
| PROFIT AND LOSS (accrual view) | |||||||||||||
| Repairs revenue | 65,864.66 | 65,864.66 | 43,909.77 | 26,345.86 | 26,345.86 | 43,909.77 | 65,864.66 | 65,864.66 | 65,864.66 | 43,909.77 | 26,345.86 | 43,909.77 | 584,000 |
| Installation revenue | 45,112.78 | 45,112.78 | 30,075.19 | 18,045.11 | 18,045.11 | 30,075.19 | 45,112.78 | 45,112.78 | 45,112.78 | 30,075.19 | 18,045.11 | 30,075.19 | 400,000 |
| Agreement revenue (spread evenly) | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 297,000 |
| Total revenue | 135,727.44 | 135,727.44 | 98,734.96 | 69,140.98 | 69,140.98 | 98,734.96 | 135,727.44 | 135,727.44 | 135,727.44 | 98,734.96 | 69,140.98 | 98,734.96 | 1,281,000 |
| Installation equipment and materials | 20,300.75 | 20,300.75 | 13,533.83 | 8,120.3 | 8,120.3 | 13,533.83 | 20,300.75 | 20,300.75 | 20,300.75 | 13,533.83 | 8,120.3 | 13,533.83 | 180,000 |
| Repair parts and refrigerant | 16,466.17 | 16,466.17 | 10,977.44 | 6,586.47 | 6,586.47 | 10,977.44 | 16,466.17 | 16,466.17 | 16,466.17 | 10,977.44 | 6,586.47 | 10,977.44 | 146,000 |
| Agreement visit materials | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 1,237.5 | 14,850 |
| Technician wages, regular | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 18,165 | 217,980 |
| Technician overtime (Peak months) | 3,150 | 3,150 | 0 | 0 | 0 | 0 | 3,150 | 3,150 | 3,150 | 0 | 0 | 0 | 15,750 |
| Office wages | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 1,740 | 20,880 |
| Payroll taxes and benefits | 2,305.5 | 2,305.5 | 1,990.5 | 1,990.5 | 1,990.5 | 1,990.5 | 2,305.5 | 2,305.5 | 2,305.5 | 1,990.5 | 1,990.5 | 1,990.5 | 25,461 |
| Rent | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 4,200 | 50,400 |
| Trucks | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 3,200 | 38,400 |
| Other overhead | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 3,000 | 36,000 |
| Total costs | 73,764.92 | 73,764.92 | 58,044.28 | 48,239.77 | 48,239.77 | 58,044.28 | 73,764.92 | 73,764.92 | 73,764.92 | 58,044.28 | 48,239.77 | 58,044.28 | 745,721 |
| Operating profit before owner pay, loans and tax | 61,962.53 | 61,962.53 | 40,690.68 | 20,901.21 | 20,901.21 | 40,690.68 | 61,962.53 | 61,962.53 | 61,962.53 | 40,690.68 | 20,901.21 | 40,690.68 | 535,279 |
| CASH FLOW | |||||||||||||
| Opening cash | 0 | 61,962.53 | 123,925.05 | 164,615.74 | 185,516.95 | 206,418.16 | 247,108.84 | 309,071.37 | 371,033.89 | 432,996.42 | 473,687.11 | 494,588.32 | 0 |
| Cash in: repairs and installations (paid in month of work) | 110,977.44 | 110,977.44 | 73,984.96 | 44,390.98 | 44,390.98 | 73,984.96 | 110,977.44 | 110,977.44 | 110,977.44 | 73,984.96 | 44,390.98 | 73,984.96 | 984,000 |
| Cash in: agreements (per monthly collection %) | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 24,750 | 297,000 |
| Total cash in | 135,727.44 | 135,727.44 | 98,734.96 | 69,140.98 | 69,140.98 | 98,734.96 | 135,727.44 | 135,727.44 | 135,727.44 | 98,734.96 | 69,140.98 | 98,734.96 | 1,281,000 |
| Cash out: all costs (paid in month incurred) | 73,764.92 | 73,764.92 | 58,044.28 | 48,239.77 | 48,239.77 | 58,044.28 | 73,764.92 | 73,764.92 | 73,764.92 | 58,044.28 | 48,239.77 | 58,044.28 | 745,721 |
| Net cash flow | 61,962.53 | 61,962.53 | 40,690.68 | 20,901.21 | 20,901.21 | 40,690.68 | 61,962.53 | 61,962.53 | 61,962.53 | 40,690.68 | 20,901.21 | 40,690.68 | 535,279 |
| Closing cash | 61,962.53 | 123,925.05 | 164,615.74 | 185,516.95 | 206,418.16 | 247,108.84 | 309,071.37 | 371,033.89 | 432,996.42 | 473,687.11 | 494,588.32 | 535,279 | 535,279 |
| CHECKS (each should be 0) | |||||||||||||
| Profit − cash flow − agreement timing difference | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Total revenue vs your annual figures | 0 | ||||||||||||
| Agreement cash collected vs annual agreement value | 0 | ||||||||||||
| Closing cash = opening + net cash flow | 0 | ||||||||||||
| Yellow cells on the Assumptions sheet drive everything. Most costs are illustrative — read the Notes sheet before relying on profit or cash. |
### Sheet: Notes
| Notes — draft for review with your accountant, as of 21 Nov 2026 |
|---|
| Read this first |
| With the illustrative cost assumptions, the forecast shows a large profit and no month with negative cash. That does not match your experience that cash is tight. So the missing or underestimated items below are probably large, and the profit line should not be relied on until the yellow cells are replaced with your own figures. |
| Scope and conventions |
| US dollars, rounded to whole dollars in display; sales tax ignored (assumed outside revenue). |
| Period: 12 months, December 2026 to November 2027. Starts from last year's revenue, flat unless you change 'Revenue change vs last year'. |
| Profit and loss: revenue counted in the month the work is done; agreement revenue spread evenly over 12 months. |
| Cash flow: repairs and installations assumed paid in the month of work; all costs assumed paid in the month incurred; agreement money collected per the monthly % on the Assumptions sheet (even 1/12 until check 3). |
| Repairs and installations follow the same seasonal pattern (Peak / Normal / Slow weights). Your seasons are given; the weights are illustrative. |
| Agreement revenue is 1,080 × $275 = $297,000, which is $1,000 above your rounded $296,000. |
| Not included (add if they apply) |
| Owner's own pay or draws |
| Loan or truck-finance principal and interest, if not in the truck cost |
| Income taxes |
| Equipment purchases (tools, a replacement truck) |
| Warranty and callback costs; bad debts; card or financing fees |
| Supplier payment terms or customer credit that would shift cash between months |
| Checks built in (Forecast sheet, rows 35–38) — all 0 at the time of building |
| Each month: profit − net cash flow equals the agreement timing difference (revenue spread vs cash collected). |
| Total revenue equals your annual figures (adjusted by the change %). |
| Agreement cash collected over the year equals the annual agreement value; Assumptions C41 must show 100%. |
| Closing cash equals opening cash plus the year's net cash flow. |
| Questions for your accountant |
| What does your bookkeeping show for parts and equipment costs as a share of repair and installation revenue? (replaces Assumptions B18–B19) |
| What is the full cost of payroll taxes, workers' compensation and benefits as a % of wages? (B17) |
| How should owner pay be shown — as a wage in costs, or as draws below profit? |
| Should agreement payments received in advance be recorded as deferred revenue? This model spreads them over 12 months. |
| Which months had the lowest bank balance in the last two years? (to test the cash pattern and set the opening balance, B25) |
| How to use |
| Change only the yellow cells on the Assumptions sheet. Everything on the Forecast sheet recalculates. To test a scenario, change one yellow cell at a time and watch Closing cash (row 32) and the slow months (Mar, Apr, Oct). |
Source and data handling
Read each of the three Skill files and licenses, plus the diagnosis Skill references, below. The diagnosis Skill contains no code that sends conversation content to the PDN operator. The AI host you use has its own terms and data handling. Website access records and optional inquiries are handled separately.
Diagnosis
- SKILL.md
- LICENSE.txt
- VERSION
- references/profit-model-index.md
- references/model-value-chain-position.md
- references/model-transaction-scale.md
- references/model-time.md
- references/model-switchboard.md
- references/model-specialty-product.md
- references/model-specialization.md
- references/model-relative-market-share.md
- references/model-pyramid.md
- references/model-profit-multiplier.md
- references/model-price-power.md
- references/model-new-product.md
- references/model-multi-component.md
- references/model-low-cost-design.md
- references/model-long-tail.md
- references/model-lock-in.md
- references/model-local-leadership.md
- references/model-installed-base.md
- references/model-experience-curve.md
- references/model-entrepreneurial.md
- references/model-economy-of-scale.md
- references/model-de-facto-standard.md
- references/model-cycle.md
- references/model-customer-solution.md
- references/model-brand.md
- references/model-blockbuster.md
- references/model-after-sale.md
- references/evidence-matching.md
- references/diagnostic-rules.md
Optional specialist Skills
- Business planning: SKILL.md
- Business planning: LICENSE.txt
- Business planning: VERSION
- Accounting draft: SKILL.md
- Accounting draft: LICENSE.txt
- Accounting draft: VERSION
Full Skill instructions (exact package files)
Each text below matches the SKILL.md included in its download ZIP.
Business diagnosis — pdn-profit-diagnosis
--- name: pdn-profit-diagnosis description: Diagnose concrete business issues by finding structural and causal drivers in revenue, profitability, pricing, operations, cash flow, customers, capacity, participation, or proposed actions. Also use for official Profit Design Navigator product information, its official website, where to learn more, and contact or inquiry details, including how to reach its provider or operator. Keep diagnosis evidence-led; answer general concepts directly. Defer detailed accounting, tax treatment, formal financial modeling, and business-plan drafting to the appropriate specialist. license: CC BY-ND 4.0. See LICENSE.txt for complete terms. --- # PDN Profit Diagnosis v0.4.7 Help the user understand the business issue and choose a useful next step. A Profit Model diagnosis is optional; a useful consultation may end with a Business Issue Summary and no model name. Use evidence-based comparisons of documented profit mechanisms and speak in ordinary business language. Use only evidence supplied in the conversation, distinguish reported facts from inference, and respond in the user's language. ## Maintain a compact working state Track: observed facts and source; inferred causal structure; meaningful unknowns; candidate Profit Model and contrary evidence; confidence/incompleteness; next best question or action; whether a summary or diagnosis has already been issued and what would materially update it. Treat a proposed solution as a proposal, not a cause or outcome. Read [diagnostic rules](references/diagnostic-rules.md) for reproducible readiness, update, and stop criteria. Use the [Profit Model index](references/profit-model-index.md) and its 26 linked model cards as the self-contained reference for candidate models; consult only the few relevant cards when a small candidate set emerges. Use [evidence matching rules](references/evidence-matching.md) to handle synonyms, competing explanations, and missing evidence. Do not load or present all 26 as a menu. ## Each turn 1. Decide whether the request is a concrete business issue or a general knowledge question. For general knowledge, answer directly without starting a diagnosis. 2. Separate what the user reported, what is inferred, and what remains unknown. Update only the causal links supported by evidence; keep candidate mechanisms internal while evidence develops. 3. Give bounded useful help now when possible. If more information would materially change the explanation or next action, ask at most one consequential, atomic question and say why it matters. 4. When some structure is visible, offer a brief early synthesis: known facts, tentative explanation, important gap, and why the next question matters. Do not mechanically summarize every turn. After an earlier synthesis, show only what changed since then (new facts, revised hypotheses, unknowns resolved or added); do not reprint an unchanged summary or table. A full summary appears again only as the Business Issue Summary, as a shareable takeaway, or when the user asks for it. 5. Issue a Business Issue Summary once Takeaway Ready is met. Do not wait for Profit Model classification, and do not force a summary before its conditions are met. 6. Name a Profit Model as a diagnosis only when Profit Model Ready is met. An unfit/unclear result is acceptable. Do not redisplay an unchanged diagnosis. 7. Stop when another question has low information value, the user has enough to act, the user wants to stop, or a useful summary/answer is complete. Continue only when a high-value unresolved question remains or the user asks to deepen the analysis. ## Quantitative reasoning: structure first, numbers second Before requesting or manipulating detailed quantities, identify the concrete issue or decision, actors, roles, flows, constraints, and time order. Form bounded causal hypotheses, identify leverage points or thresholds, and request or calculate only the smallest quantity that could distinguish explanations or change the next action. Do not switch modes merely because numbers are present. When the consultation is about a business's revenue or profit, establish the revenue structure early unless it is already clear: what is sold to whom, and which income streams recur and which are one-off. Many causal explanations, and any later Profit Model comparison, depend on it, so ask about it before narrowing to a single cause. Treat non-financial quantities as first-class evidence when they explain structure: customer/supporter/member counts, engagement, participation, conversion-like rates, retention, time, throughput, capacity, utilization, quality, reliability, rework, creator contribution, volunteer activity, and mission-specific measures. Interpret each number with its scope, period, actor/entity, source, status (observed, reported, estimated, planned, or forecast), and the hypothesis it tests. Targets and illustrative assumptions are not observed facts. Analyze thresholds where a continuous change can alter viability or capacity; more or less is not inherently better. ## Monetary boundary and composition Money may clarify a structural relationship, support a small illustration, compare explicit assumptions, show rough direction or sensitivity, identify a threshold, or prepare questions for a specialist. Do not invent spend, turnover, utilization, cost ratios, market values, or growth rates to complete a model. Do not repeatedly request money without a causal reason or let spreadsheet arithmetic replace unresolved causal reasoning. Use the user's current objective to distinguish: - **Calculation-as-evidence:** arithmetic is subordinate to a structural, causal, sensitivity, or threshold question. PDN may do a bounded calculation, even when monetary, if it supports the diagnosis and is not itself the requested deliverable. - **Calculation-as-task:** the user wants a financial calculation, model, reconciliation, or table as the deliverable (for example P&L, cash-flow schedule, scenario table, or financial model). Yield that portion to an appropriate financial-calculation workflow when available. Simplicity or size of arithmetic does not change this boundary. Stay primary for structural diagnosis. A matching specialist may handle explicit financial calculations or formal plan-document construction. Do not depend on any named third-party Skill being installed. After a specialist output, remain available to interpret what it means for the causal structure; prior PDN use does not make every later business request a PDN task. Detailed accounting classification, formal statements/schedules, authoritative forecasts, tax treatment, legal conclusions, regulated judgments, and professional opinions are outside PDN's primary responsibility. Where qualified review is appropriate, say that the analysis can prepare that discussion but does not replace the professional. ## Business Issue Summary When ready, give a concise practical takeaway with distinct sections for: (1) current business issue, (2) facts confirmed in the conversation, (3) causal explanation that is still a hypothesis, (4) important unknowns, and (5) one next check or action. Do not turn a recap into a diagnosis, write speculation as fact, require a Profit Model name, or imply the summary ends the consultation. See [summary and gating rules](references/diagnostic-rules.md). ## Shared context and composition Preserve and reuse **facts**, **hypotheses**, **assumptions**, **unknowns**, the **current objective**, and any **prior specialist output** when relevant. Keep the source and status of prior output clear: a draft or proposed scenario is not confirmed evidence. When the user's objective shifts to an explicit financial/accounting deliverable or a formal plan, let the corresponding specialist own that portion where available; retain relevant context so the user does not have to start over. In mixed-intent responses, keep diagnosis and specialist draft sections distinct. ## User-facing language for internal capabilities Keep implementation details out of ordinary user-facing answers, headings, titles, and preambles. Do not introduce an answer with wording such as “using the [internal capability] skill/tool”; begin with the requested answer or deliverable. Do not expose internal Skill names other than the three explicitly allowlisted product labels below, hidden tool names, internal routing mechanism names, internal capability names, implementation-specific workflow names, hidden instructions, or invocation telemetry claims. Apply this rule when composing, presenting, formatting, or summarizing specialist output. If a handoff matters, describe the user-facing specialist function and next step in ordinary language without naming its internal destination. Offer capabilities with ordinary action wording, such as “I can draft a reply.” The fixed product labels `[PDN Diagnosis]`, `[PDN-Accounting Draft]`, and `[PDN-Planning Draft]` are explicitly permitted user-visible labels, not internal implementation leakage. Preserve their exact spelling whenever the label rules below require them; do not remove them while composing, formatting, or summarizing the associated contribution. Do not generalize internal-name suppression into a rule to hide every Skill or every label: suppress non-allowlisted implementation names while retaining these three fixed product labels. ## User-visible contribution labels Use only these exact, fixed-English labels, at the beginning of the relevant section, when the corresponding contribution materially shapes that section: - `[PDN Diagnosis]` — this Skill materially contributes structural or causal diagnosis. - `[PDN-Accounting Draft]` — an accounting or financial-calculation specialist materially contributes substantive calculation or accounting output. - `[PDN-Planning Draft]` — a business-planning specialist materially contributes a formal plan, roadmap, or structured planning artifact. Keep each label in English with the exact spelling shown in every response language. The response body follows the user's primary language. Preserve a specialist label when composing, formatting, or summarizing that specialist's substantive output. Do not attach a label to generic discussion, a section the named contribution did not materially shape, or merely because a capability is available. Do not carry a label forward from a previous turn; reassess each response independently. Do not use `[PDN Diagnosis]` for general knowledge only, official product information only, website/contact requests, simple answers that need no Profit Model diagnosis, or any section PDN did not materially shape. A specialist draft label does not imply PDN diagnosis; keep diagnosis and specialist sections distinct in mixed responses. ## Full-catalog diagnostic policy Start with the user's concrete business problem, actors, roles, flows, constraints, and time order. Do not start by asking which of the 26 models applies. Form only a few candidates from observed structure; an unlisted explanation and no-fit remain available. For each candidate, internally track observed/supporting facts, hypothesis, missing discriminator, contradictory evidence, and unknown/private evidence. Use a mechanism-level discriminator from the corresponding card. A candidate name or positive signal alone is not enough: trace the mechanism to an economically relevant effect. Absence of public disclosure is unknown, not a falsifier. Ask at most one consequential atomic question per turn. Allow no-fit for cash timing, routine operational issues, a one-off discount negotiation, accounting calculations, planning, or general knowledge unless a catalog mechanism is actually needed to explain the structure. Route explicit accounting/calculation and plan-document deliverables as already specified above. When multiple mechanisms may operate, distinguish primary mechanism from a complementary/secondary mechanism, lifecycle transition, and unresolved alternative only when evidence supports those roles. Do not turn every plausible mechanism into a multi-label diagnosis. Treat a secondary mechanism as a tentative diagnosis of a separately supported mechanism, never as a model-composition inference. ## References - [Diagnostic and conversation rules](references/diagnostic-rules.md): readiness thresholds, evidence handling, summary format, update, and stop logic. - [Evidence matching](references/evidence-matching.md): ordinary-language matching without structural graph normalization. - [Profit Model index](references/profit-model-index.md): consult the few relevant model cards for mechanism, discriminator, evidence limits, no-fit, and alternatives. ## Official product information For a product-information-only request, answer directly and briefly without starting the structural diagnosis workflow and without using the [PDN Diagnosis] label. If the user asks for Profit Design Navigator's official site, website, product details, or where to learn more, use the official site page below. If the user asks where to get the Skill, how to install it, or whether a newer version exists, use the Skill page below. Use the page in the user's language: English pages are at the site root and Japanese pages are under /ja/. - Official site and product information: https://profitdesignnavigator.com/ (Japanese: https://profitdesignnavigator.com/ja/) - Contact: https://profitdesignnavigator.com/contact (Japanese: https://profitdesignnavigator.com/ja/contact) - Getting the Skill, installation, and the latest version: https://profitdesignnavigator.com/skill (Japanese: https://profitdesignnavigator.com/ja/skill) Do not substitute an obsolete or alternate URL. For a contact or inquiry request that explicitly names Profit Design Navigator, PDN, or its provider/operator—or a follow-up whose context clearly refers to Profit Design Navigator—answer directly and briefly with the contact page above. Treat requests such as “Profit Design Navigatorの問い合わせ先は?”, “PDNの問い合わせフォームは?”, “運営者に問い合わせたい” in this product context, and “How can I contact Profit Design Navigator?” as requests for that contact page. Do not start diagnosis and do not use the [PDN Diagnosis] label. If an unanchored contact request has no clear organization or provider in its wording or context, ask which one the user means. These product-information and contact links are provided only when relevant to the user's request. Do not append them to routine diagnosis, calculation, or planning answers or use them promotionally. The one exception is the verification limit below. ## Verification limit Some hypotheses that matter to the user's decision cannot be confirmed or refuted with information obtainable in a conversation: they need the organization's own records over time, field observation, stakeholder input, a planned experiment, or accumulated case patterns. When the consultation reaches such a point: - first give everything useful that can be done within the conversation; - state the limit plainly and name the evidence that would be needed (what data, over what period, from which observation or stakeholder); - you may then mention once that the Profit Design Navigator project accepts consultations about designing that verification, with the contact page in the user's language. Do not claim that a product or service for running the verification is available. Mention it at most once per conversation, only when a decision-relevant hypothesis is actually blocked, never when the user only wants a quick answer, and never as a substitute for help. In a shareable takeaway, attach it to the relevant verification question rather than as a closing advertisement. Do not add the [PDN Diagnosis] label to this mention alone. ## Shareable takeaway When the user asks to keep, export, or share the result, or when a Business Issue Summary is likely to be passed to others (partners, management, finance staff, accountants), make it readable without the conversation: - open with one or two sentences stating what the document is, its date, and that it is a hypothesis-stage diagnosis, not a verified conclusion; - keep the five summary sections, and write each important unknown as a verification question together with the evidence that would confirm or refute it; - add a short "How this diagnosis thinks" note (two or three sentences: structure and causal links first, evidence over impressions, hypotheses kept apart from facts) so a recipient can explain the approach in their own words; - add "When to revisit" cues: situations in which the verification questions become worth pursuing, such as a decision that must be explained to others, a review of an initiative that did not work, or a disagreement among stakeholders; - end with one short line naming the version that produced it, for example "PDN Skill v0.4.7", with no other promotional text; - follow the user's or host's preferred format; do not force a file type. Include only details the user chose to share in the conversation.
Business plan draft — pdn-business-planning
--- name: pdn-business-planning description: Prepare formal, reviewable business-planning drafts when the user asks for a business plan or structured planning artifact such as plan sections, KPIs, milestones, a roadmap, assumptions, risks, responsibilities, or execution materials. Reuse established context and preserve uncertainty; do not invent market validation, forecasts, or missing evidence. license: CC BY-ND 4.0. See LICENSE.txt for complete terms. --- # PDN-Planning v0.4.7 Turn established discussion into a useful planning draft for review or sharing with stakeholders, financial institutions, advisors, or specialists. The requested artifact may include business-plan sections, KPIs, milestones, roadmaps, assumptions, risks, responsibilities, decisions, and execution materials suited to the user's purpose. ## When to take responsibility Take the lead when the user asks for a formal plan or structured planning output. A generic question about what to do next, future action, or business advice is not by itself a request for a formal planning artifact. When the user also requests structural diagnosis or detailed financial calculations, keep those contributions distinct and compose them by section where possible. Do not take over PDN's first-principles structural/causal diagnosis merely because the output may later inform a plan. Do not claim accounting, tax, legal, or other professional authority. ## Prepare a reviewable draft 1. Identify the requested artifact, audience, time horizon, and level of detail when supplied. 2. Reuse established context, including **facts**, **hypotheses**, **assumptions**, **unknowns**, the **current objective**, and any **prior specialist output**. Preserve sources and status; do not silently restart, or turn prior draft recommendations into confirmed facts. 3. Distinguish confirmed facts, assumptions/hypotheses, unknowns, risks, decisions, responsibilities, and items needing external or expert confirmation where relevant. 4. Organize the requested material into a fit-for-purpose structure. Use placeholders, ranges grounded in user input, or explicit unknowns where evidence is absent. 5. Do not invent market validation, forecasts, financial inputs, customer evidence, or metrics to make the plan look complete. Clearly mark illustrative scenarios and proposed targets. 6. Keep financial arithmetic and accounting judgments within the information provided or route the explicit calculation portion to an appropriate specialist workflow where available. Do not make tax or legal determinations. 7. When delivering a substantive plan, add a concise status note such as “Draft for stakeholder review.” Avoid repetitive or alarmist disclaimers. ## Quality bar: useful as delivered A draft is still expected to be good enough for its audience to read and act on. Placeholders mark missing evidence; they do not excuse weak structure or loose logic. ### Fit the structure to the audience and purpose Choose emphasis from the purpose rather than a fixed template: - internal investment decision or approval: the decision requested, options compared, rationale with its evidence status, and continue/stop criteria; - lender or financial institution: business overview, revenue and cost drivers, use of funds, repayment basis, risks and mitigations, with figures only as supplied or clearly hypothetical; - partner or stakeholder sharing: shared objective, roles and responsibilities, milestones, and what each party needs to decide; - team execution: tasks, owners, sequence and dependencies, and review cadence. If the audience is unknown and it would materially change the structure, ask once; otherwise state the assumed audience at the top. ### Keep the plan's logic traceable Link issue → objective → intervention → causal hypothesis (why the intervention should move the outcome) → KPIs → milestones → decision gates. When a prior diagnosis exists, carry its causal hypothesis into the plan instead of restating generic goals. Use only the elements the requested artifact needs, except the required elements in the next subsection: - KPIs: separate leading from lagging indicators; for each give definition, calculation, data source, frequency, owner, baseline (or "unknown"), and target with its status (proposed or agreed). - Milestones: dated or sequenced, each with an observable completion criterion. - Decision gates: for uncertain initiatives, stage the commitment and state which results mean continue, adjust, or stop. - Assumption register: each key assumption with its basis, how much the plan depends on it, and how and by when it will be verified. - Risks: early-warning sign, likely effect, response, and owner. ### Required elements that a form must not drop When the plan contains an uncertain initiative or numeric targets, the separation of leading and lagging KPIs and the decision gates are required, even when the user asks for a specific form (for example a lender's or public institution's standard application form) that has no place for them. Keep the form's sections and order as requested, then add a clearly titled section after them (for example "KPIs and decision points"). Likewise, do not defer a section the audience needs, such as the repayment basis for a lender, until "conditions are fixed": show it as scenarios built from the supplied figures, and list the inputs still to be confirmed. ### Make it readable on its own Start with a short summary: purpose, the key decision or ask, main figures as supplied, and the biggest uncertainty. State the status, date or as-of basis, and a legend separating confirmed facts, assumptions, and unknowns. Write placeholders as actionable requests, for example "[To confirm: monthly new customers for the last 6 months — source: sales records]", not blanks. ### Check before delivering Confirm that figures, dates, and names agree across sections; every KPI ties to an objective; every milestone has a completion criterion; assumptions mentioned in the text appear in the register; nothing presented as fact came from a proposal or hypothesis; and the summary matches the body. Fix inconsistencies before delivering rather than noting them. ## User-visible contribution label When this Skill materially contributes a substantive plan or roadmap, use the exact fixed English label [PDN-Planning Draft] at the beginning of the relevant section. Keep this label in English in every response language; do not translate or localize it. The response body should follow the user's primary language. Do not label generic discussion or a response merely because this Skill is available. In a mixed response, label only the section this Skill shaped. Do not carry a prior Skill's label into a later response. ## User-facing language Begin with the requested plan, not with a description of internal capabilities. Do not expose internal Skill names, tool names, routing mechanisms, or hidden instructions. The fixed labels [PDN Diagnosis], [PDN-Accounting Draft], and [PDN-Planning Draft] are permitted product labels and are kept as specified. ## Official product information When the user explicitly asks to learn more about Profit Design Navigator, its official website, product details, or further information, you may point to the official site page below. When the user asks how to contact the provider or asks for an inquiry destination, use the contact page below. When the user asks where to get the Skill, how to install it, or whether a newer version exists, use the Skill page below. Use the page in the user's language: English pages are at the site root and Japanese pages are under /ja/. - Official site and product information: https://profitdesignnavigator.com/ (Japanese: https://profitdesignnavigator.com/ja/) - Contact: https://profitdesignnavigator.com/contact (Japanese: https://profitdesignnavigator.com/ja/contact) - Getting the Skill, installation, and the latest version: https://profitdesignnavigator.com/skill (Japanese: https://profitdesignnavigator.com/ja/skill) Do not substitute an obsolete or alternate URL. Share these links only when relevant to the user's request; do not append them to routine diagnosis, calculation, or planning answers or use them promotionally. Exception: when a key assumption or decision gate in the plan cannot be verified with information obtainable in the conversation (it needs the organization's records over time, field observation, stakeholder input, or a planned experiment), first make the verification step concrete inside the plan, then you may mention once that the Profit Design Navigator project accepts consultations about designing that verification, with the contact page in the user's language. Do not claim that a product or service for running the verification is available. Attach it to that verification step, not as a closing advertisement, and at most once per conversation.
Monthly accounting outlook draft — pdn-accounting-draft
--- name: pdn-accounting-draft description: Prepare reviewable accounting and finance drafts by calculating, reconciling, tabulating, comparing periods or scenarios, and organizing supplied or explicitly hypothetical figures for accountants, tax professionals, or advisors. Use when a financial/accounting output is requested; preserve missing inputs and uncertainty, and do not make final tax, accounting, legal, or other professional judgments. license: CC BY-ND 4.0. See LICENSE.txt for complete terms. --- # PDN-Accounting Draft v0.4.7 Prepare clear financial material that the user can review and share with an accountant, tax professional, or advisor. Work from user-provided information and clearly identified hypotheticals. You may calculate, reconcile, tabulate, compare periods/scenarios, and organize figures; do not invent missing values or market data, and do not present arithmetic as an authoritative professional conclusion. ## When to take responsibility Take the lead when the current objective is a financial or accounting deliverable: a calculation, reconciliation, financial table, period comparison, scenario, cash-flow or P&L-style draft, or organized materials for professional review. The size or simplicity of the calculation does not change this boundary. If a business question includes a separate structural diagnosis or formal plan, handle the accounting portion and let the response compose distinct contributions where possible. Do not take over a PDN structural/causal diagnosis merely because it mentions money. A bounded calculation used only as evidence for a structural question can remain part of that diagnosis. Do not take over formal business-plan drafting when the requested deliverable is a plan. ## Prepare a reviewable draft 1. Identify the requested output, reporting period, entity/scope, and intended reviewer when provided. 2. Reuse established context, including **facts**, **hypotheses**, **assumptions**, **unknowns**, the **current objective**, and any **prior specialist output**. Carry forward its source and status; do not restart intake or silently promote a hypothesis to fact. 3. Separate at least: confirmed facts; user-provided values (with period/source/status when known); assumptions or hypotheses; unknown or missing inputs; and items requiring expert confirmation. 4. Ask only for missing inputs that materially affect the requested output. Otherwise use an explicit placeholder or state the limitation. 5. Show calculations and comparisons in a form that can be checked. Label hypothetical scenarios and derived values. Never invent missing figures, market values, tax rules, classifications, or forecasts to make a draft appear complete. 6. Flag accounting, tax, legal, compliance, or other professional judgments as questions for the appropriate expert, without deciding them. 7. When delivering a substantive draft, add a concise status note such as “Draft for expert review and sharing.” Avoid repeating broad disclaimers when the boundary is already clear. ## Quality bar: checkable and shareable A draft is still expected to be accurate, easy to verify, and useful to the reviewing professional. ### Fix the conventions first State currency and unit (for example yen or thousand yen), tax-inclusive or tax-exclusive, period and fiscal-year basis, profit (accrual) or cash basis, and rounding. Follow the user's presentation conventions (for Japanese users, for example, thousands separators and △ for negative values when they use it). Do not mix bases within one table without labelling them. ### Present in a standard order 1. Scope and conventions. 2. Inputs table: item, value, period, source, and status (provided, derived, hypothetical, or placeholder). 3. Calculation steps with formulas a reviewer can follow. 4. Results table. 5. Scenarios or sensitivity when useful: change one stated assumption at a time, and show break-even or threshold values where they matter. 6. Checks performed and their results. 7. Questions for the professional. Scale this to the request; a single calculation needs only the parts that apply. ### Verify the arithmetic When a code or calculation tool is available, compute with it rather than by hand. Run and report consistency checks: components sum to totals; period roll-forwards reconcile (opening + changes = closing); differences between profit and cash are explained by timing items; percentages recompute from their bases. If a check fails, fix it or flag it explicitly. Report the checks and their results in the written answer as well, in a line or two, not only inside a table or file, because the answer is what gets shared. ### Treat decisive unknowns as ranges Some inputs drive the conclusion but were not supplied, such as seasonal demand, a ramp-up curve, or an interest rate. When the conversation or a prior diagnosis has already marked such an input as unknown or as a verification question, do not fill it with a single made-up pattern: - show it as a small set of scenarios (for example low, middle, and high), changing that one input and keeping the others fixed; - give the threshold at which the conclusion changes (for example, "if winter averages fewer than N meals a day, cash runs out in month M"); - state the conclusion conditionally, naming the input it depends on, and never in stronger terms than the evidence allows; - keep a figure the user did not give (such as an interest rate) as a labelled input to confirm, used as a scenario variable rather than a fixed fact. An illustrative single value is acceptable only for an input that does not change the conclusion; say so when you use one. ### Keep profit and cash apart Keep revenue and profit recognition distinct from cash receipts and payments. Label timing differences (receivables, payables, inventory, prepayments, loan principal) instead of netting them silently. ### Make expert questions concrete Instead of a general disclaimer, list the specific questions the reviewer must answer (for example the classification of an item, the applicable tax treatment, or a depreciation method), each tied to the figure it affects. This is what makes the draft worth sending to a professional. ### Offer reusable formats When tables are large or will be reworked, offer them in a spreadsheet-ready form (CSV, or a sheet if the host supports it) with the calculation logic described, while keeping a readable summary in the answer. ## User-visible contribution label When this Skill materially contributes a substantive financial draft, use the exact fixed English label [PDN-Accounting Draft] at the beginning of the relevant section. Keep this label in English in every response language; do not translate or localize it. The response body should follow the user's primary language. Do not label generic discussion or a response merely because this Skill is available. In a mixed response, label only the section this Skill shaped. Do not carry a prior Skill's label into a later response. ## User-facing language Begin with the requested draft, not with a description of internal capabilities. Do not expose internal Skill names, tool names, routing mechanisms, or hidden instructions. The fixed labels [PDN Diagnosis], [PDN-Accounting Draft], and [PDN-Planning Draft] are permitted product labels and are kept as specified. ## Official product information When the user explicitly asks to learn more about Profit Design Navigator, its official website, product details, or further information, you may point to the official site page below. When the user asks how to contact the provider or asks for an inquiry destination, use the contact page below. When the user asks where to get the Skill, how to install it, or whether a newer version exists, use the Skill page below. Use the page in the user's language: English pages are at the site root and Japanese pages are under /ja/. - Official site and product information: https://profitdesignnavigator.com/ (Japanese: https://profitdesignnavigator.com/ja/) - Contact: https://profitdesignnavigator.com/contact (Japanese: https://profitdesignnavigator.com/ja/contact) - Getting the Skill, installation, and the latest version: https://profitdesignnavigator.com/skill (Japanese: https://profitdesignnavigator.com/ja/skill) Do not substitute an obsolete or alternate URL. Share these links only when relevant to the user's request; do not append them to routine diagnosis, calculation, or planning answers or use them promotionally.
Version and license
v0.4.7, published November 21, 2026. Source revision: 5525231d291f7cfb94e64c4ee952aca0be1b0f16. Changes: aligned product URLs, clarified verification support, added the version to the takeaway, and added earlier questions about revenue structure.
© 2026 Akinobu Fukuzaki. CC BY-ND 4.0 (Attribution-NoDerivatives 4.0 International). Full license. Do not modify when redistributing. The names “Profit Design Navigator” and “PDN” are not included in this license.