Plan

Plan — Versioned Strategies for Goals

Overview

Plans are the tactical layer on top of goals. A goal is stable: it declares what you want and when you want it. A plan is a versioned strategy: it says how much to contribute per month, from which sources, and in what order if multiple goals compete. Plans iterate as life changes — a raise, a new obligation, a market drop all warrant a new plan version. The old plan is marked superseded, never deleted. Goals are stable; plans iterate.

Locate the workspace: read ~/.coinskills-workspace for the absolute path. If the file doesn't exist, stop and tell the user to run /coinskills:init first.

---

Step 1: Select Goal(s)

Scan goals/*.md in the workspace. Extract frontmatter from each file. List only goals with status: active.

Print the active goals table:

| id | title | type | target | deadline | priority | latest plan |
|----|-------|------|--------|----------|----------|-------------|
| house-deposit | House deposit | savings | €50,000 | 2028-06-01 | 1 | v2 |
| emergency-fund | Emergency fund | savings | €12,000 | none | 2 | — |

For latest plan version: scan plans/<id>-v*.md — find the highest version number among files with status: active. If none exists, show —.

Ask the user: "Which goal(s) do you want to plan for? (enter one id or a comma-separated list — multi-goal plans are allowed)"

Wait for the answer. Store the selected goal id(s) as selected_goals. If the user names a goal not in the active list, inform them and ask again.

If goals/ is empty or no goals have status: active, tell the user they have no active goals and suggest /coinskills:goals new.

---

Step 2: Analyze Current State

Load the following files from the workspace:

• profile.md — currency, emergency_fund_months, risk_tolerance, locale, modules
• accounts.json — all accounts with type, balance, apr, billing_cycle_day, limit
• modules/personal/recurring.json — fixed recurring obligations (subscriptions, rent, utilities, loan repayments)
• modules/personal/income.json — income streams (salary, freelance, rental, dividends) with amounts and frequencies
• modules/personal/transactions/YYYY-MM.md for the last 3 calendar months (e.g. if today is 2026-04-27, load 2026-02, 2026-03, 2026-04)

If a file doesn't exist (e.g. transactions not yet logged), note it and proceed with the data that is available. If income.json is missing, ask the user for monthly income before proceeding.

Compute monthly contribution capacity:

monthly_income         = sum of all income.json streams normalised to monthly frequency
                          (weekly * 52 / 12, annual / 12, etc.)

recurring_obligations  = sum of recurring.json amounts normalised to monthly
                          (include only obligations that are NOT goal contributions)

essential_spending_baseline =
    if 3 months of transactions exist:
        avg of last 3 months' total outflows, EXCLUDING:
        - rows with category "goal-contribution"
        - rows with category "investment-buy" or "investment-sell"
        - one-off large purchases (any single row > 5x the per-category monthly average)
    else:
        recurring_obligations  (conservative fallback — note this to the user)

monthly_contribution_capacity =
    monthly_income - recurring_obligations - essential_spending_baseline

Print a summary:

Monthly income:            €{monthly_income}
Recurring obligations:     €{recurring_obligations}
Essential spending:        €{essential_spending_baseline}
---
Available for goals:       €{monthly_contribution_capacity}

If monthly_contribution_capacity is zero or negative, warn the user: "Your obligations and spending currently leave nothing for contributions. We can still model a plan — but it will show a very long timeline unless something changes." Proceed anyway; the what-if scenarios in Step 4 may surface a path forward.

---

Step 3: Propose Contribution Schedule

For each goal in selected_goals, propose a monthly contribution based on goal type.

savings / purchase / retirement

1. Compute months remaining: (deadline - today) in months, or ask the user for a target horizon if deadline is none.
2. Required monthly contribution: (target_amount - current_saved) / months_remaining
   • current_saved = sum of balances of linked_accounts from accounts.json where account type is bank, savings, or e_money. If no accounts are linked, ask the user for the current saved amount.
3. Compare required vs available capacity. If required > capacity, show both numbers and say: "At your current capacity you'd reach this goal in {N} months rather than {deadline}. The what-if scenarios below may help."
4. Ask: "Which income sources should fund this goal? (e.g. salary, freelance, side-income — or just 'all')"
5. Build a contribution_sources list: split the monthly amount proportionally across the named sources by their weight in income.json. If the user says all, use all streams proportionally.

Print:

Goal: {title}
  Monthly contribution needed:  €{required}
  Monthly contribution feasible: €{min(required, capacity)}
  Projected completion:          {date}
  Sources:
    {account}: €{amount}/month
    ...

debt-payoff

For each debt account in accounts.json (types: credit_card, loan, mortgage) that is linked to this goal:

1. Collect: balance (as positive), APR, minimum payment.

2. Present two strategies and recommend one with a brief justification:

   • Snowball (smallest balance first): pays off individual debts faster, gives psychological momentum. Recommended when the user has multiple small debts and motivation is a factor, or when APRs are similar (within 3 percentage points of each other).
   • Avalanche (highest APR first): minimises total interest paid. Recommended when there is a significant APR spread (>3pp) — mathematically optimal. Default recommendation unless snowball criteria apply.

3. State the recommendation clearly: "I recommend avalanche because the APR spread between your debts is {N}pp — over {timeline} this saves you approximately €{savings} in interest compared to snowball." Or: "I recommend snowball because your debts have similar APRs ({range}%) and you have {N} small accounts — the quick wins will help you stay on track."

4. Under the chosen strategy, schedule payments: minimum payments on all other debts, extra capacity goes entirely to the priority debt.

Print the payment schedule:

Strategy: Avalanche (highest APR first)
Rationale: {rationale}

Payment order:
  1. {debt-id} — APR {apr}% — balance €{balance} — payoff by {date}
  2. {debt-id} — APR {apr}% — balance €{balance} — payoff by {date}

Monthly allocation: €{total}
  Minimums:   €{sum_minimums}
  Extra:      €{extra} → {priority-debt-id}

investment

1. Ask: "What is your target allocation? (e.g. 70% equities / 20% bonds / 10% cash — or 'from profile' to use risk_tolerance defaults)"
   • If from profile: conservative → 40/40/20, moderate → 60/30/10, aggressive → 80/15/5.
2. Propose a monthly DCA (dollar-cost averaging) amount = min(required, capacity).
3. Note any allocation drift: compare current holdings in modules/investments/holdings.json (if module enabled) to the target. If drift > 5pp on any asset class, surface it: "Your current allocation is {actual}% equities vs target {target}% — consider rebalancing as part of this plan."

Print:

Goal: {title}
  Monthly DCA: €{amount}
  Target allocation: {equities}% equities / {bonds}% bonds / {cash}% cash
  Current drift: {note or "within tolerance"}
  Projected portfolio value at {deadline}: €{estimate} (assumes {rate}% annual return)

For the projection, use 7% annual return for equities-heavy allocations (>60% equities), 4% for balanced, 2% for conservative. State the assumption explicitly.

Multi-goal plans

If selected_goals contains more than one goal, allocate monthly_contribution_capacity across goals weighted by priority (priority 1 gets the largest share). Print the allocation table:

Total available: €{capacity}
  {goal-id-1} (priority 1): €{amount1}/month
  {goal-id-2} (priority 2): €{amount2}/month

Ask the user: "Does this allocation look right, or would you like to adjust the split?" Wait for confirmation or adjustment before proceeding.

---

Step 4: What-If Scenarios

Tell the user: "I can model three what-if scenarios. Each shows how your projected completion date changes. Run all three, pick some, or skip."

Present the options:

1. What if I get a 10% raise?
2. What if a recession halves my freelance income for 6 months?
3. What if I take on a new €X/month obligation?

For option 3, ask: "What monthly amount for the new obligation?" before modelling.

Ask: "Which what-ifs would you like to see? (1, 2, 3, any combination, or 'all' / 'skip')"

For each selected scenario, recompute monthly_contribution_capacity with the changed assumption and print a revised projected completion date for each goal in selected_goals:

Scenario 1 — 10% raise:

New monthly income:        €{monthly_income * 1.10}
New capacity:              €{new_capacity}
Revised projections:
  {goal-id}: {old_date} → {new_date} ({delta} earlier)

Scenario 2 — Recession halves freelance for 6 months:

Freelance income for 6 months: €{freelance * 0.5}/month (then restores)
Reduced capacity for 6mo:      €{reduced_capacity}
Normal capacity from month 7:  €{normal_capacity}
Revised projections:
  {goal-id}: {old_date} → {new_date} ({delta} later)

Scenario 3 — New €X/month obligation:

New obligation:            €{X}/month
New capacity:              €{capacity - X}
Revised projections:
  {goal-id}: {old_date} → {new_date} ({delta} later)

If there is no freelance income stream in income.json, for Scenario 2 substitute the largest non-salary income stream, or if all income is salary, say "You have no variable income — a recession would not change your freelance income, but job loss would. Modelling 6-month salary reduction to 50% instead." and proceed with that assumption.

---

Step 5: Versioning

For each goal id in selected_goals:

1. Find existing active plan files. Scan plans/{goal-id}-v*.md. For each file found, read the status frontmatter field. If status: active, mark it superseded:

   • Open the file.
   • Change the frontmatter field status: active to status: superseded.
   • Save the file.
   • Note the old version number (e.g. v2).

2. Determine the new version number. Find the highest version number across ALL plan files for this goal (active, superseded, or abandoned). Increment by 1. If no prior plans exist, start at v1.

3. Write the new plan file at plans/{goal-id}-v{N+1}.md using this exact template:

---
goal_ids: [house-deposit]
version: 1
created: 2026-04-27
status: active
monthly_contribution: 1400
contribution_sources:
  - {account: salary, amount: 900, frequency: monthly}
  - {account: freelance, amount: 500, frequency: monthly}
projection: on-track
---
# Strategy
Narrative: how this plan works, assumptions, what triggers a v2.

Fill in the fields:

• goal_ids: list all ids from selected_goals
• version: the new version number (integer)
• created: today's date (YYYY-MM-DD)
• status: active
• monthly_contribution: total monthly contribution from Step 3
• contribution_sources: list from Step 3, one entry per source with account, amount, frequency
• projection: on-track if required ≤ capacity, behind if required > capacity

For the # Strategy body, write a 2-4 sentence narrative covering:

• What the plan does (monthly amount, sources, method for debt-payoff goals)
• Key assumptions (income stability, APR assumptions, return rate for investments)
• What would trigger a revision (a raise, a new obligation, a missed contribution for 2+ months)

For multi-goal plans (when selected_goals has more than one id), write a single plan file covering all goals. Name the file using the first goal id: plans/{first-goal-id}-v{N+1}.md. Set goal_ids to the full list.

4. Tell the user what happened:
   • "Marked prior plan(s) as superseded: {list of filenames}"
   • "Wrote new plan: plans/{goal-id}-v{N+1}.md"

---

Step 6: Update Goal Projection

For each goal id in selected_goals:

1. Read goals/{goal-id}.md.
2. Check if the frontmatter contains a projection field. If it does, recompute it:
   • If the new plan's projected completion date ≤ deadline: set projection: on-track
   • If projected completion date > deadline by ≤ 3 months: set projection: behind
   • If projected completion date > deadline by > 3 months: set projection: at-risk
   • If projected completion date < deadline by ≥ 1 month: set projection: ahead
3. If the projection field does not exist in the frontmatter, add it.
4. Save the file.

Tell the user: "Updated projection for {goal-id}: {projection value}."

---

Commit

After all files are written and goals updated, stage and commit from the workspace root:

git add goals/ plans/
git commit -m "plan: {goal-id} v{N} ({trigger reason})"

Where {trigger reason} is a short phrase describing why a new plan was created. Derive it from context:

• First plan ever → initial
• User mentioned a raise → raise factored in
• User mentioned a new obligation → new obligation: {description}
• User mentioned changed income → income revised
• Plan revised after goal edit → goal {id} revised
• No explicit reason given → revised

Examples:

• plan: house-deposit v3 (raise factored in)
• plan: house-deposit v1 (initial)
• plan: emergency-fund v2 (new obligation: car loan)
• plan: house-deposit emergency-fund v2 (income revised)

For multi-goal plans, list the goal ids space-separated before the version: plan: {goal-id-1} {goal-id-2} v{N} ({reason}).

---

Self-check before finishing

Before declaring done, verify:

• Every goal in selected_goals has a new plan file written at plans/{goal-id}-v{N+1}.md with status: active
• All previously active plans for those goals now have status: superseded
• The projection field in each goals/{goal-id}.md reflects the new plan
• The commit has been run and confirmed