Canadian Personal Finance Planner Skill
Turns Claude into a personal financial coach that knows Canadian taxes and benefits: interviews, budgets, full plans, an interactive dashboard, and ongoing coaching.
Evidence shows explicit sensitive-data handling (SINs, account and card numbers fully ignored, never echoed), local-only file storage, and clear privacy disclosure with training opt-out guidance. Deducted for: multi-file writes to user folders plus recommended python dependencies (markdown2/weasyprint/openpyxl) without version pinning or supply-chain disclosure; no rollback/deletion mechanism; Anthropic server claims are asserted, not verifiably attributed.
Evidence shows multi-stage validation (data completeness, recalculation from scratch, cross-file consistency), progressive on-demand reference loading, and a self-consistent architecture. Deducted for: static review cannot execute paths; SKILL.md text is truncated mid-section ('Handling Edge Cases'), leaving instructions incomplete; no committed tests or CI covering key paths.
Evidence shows a clear Canadian-resident audience with an explicit 'Canadian Gate' declining non-Canadian users, well-defined trigger phrases, and declared non-fit boundaries (no tax filing, no stock picks, not legal advice). Deducted for: English only; core function depends on paid Claude plans and live web search, raising mainland-China reachability concerns; reference file contents not reviewed in this evidence set.
Evidence shows MIT license, version and date stamps, changelog links, a well-layered doc structure (SKILL.md plus 17 on-demand reference files), and a maintainer feedback channel. Deducted for: unverified publisher with personal maintenance responsibility; truncated SKILL.md; dependency setup instructions split off into README rather than the skill.
Evidence shows complete output definitions (budget, plan, interactive dashboard, PDF/Excel) and claims of a demo sample plus 12 test personas with a scoring rubric. Deducted for: static review cannot verify output usability; PDF/Excel generation depends on external libraries that may be absent, degrading outputs; no executed verification of representative outputs.
Evidence shows 150+ claimed source URLs, a two-source verification process, demo-sample-plan and test-scenarios folders with a rubric, and key citable sources listed in README. Deducted for: the reference files themselves are not in the evidence set, so verification dates and figures cannot be independently checked; no third-party execution evidence.
- SKILL.md is truncated at 'Handling Edge Cases', so instructions are incomplete and actual behavior may diverge from documentation.
- PDF/Excel generation depends on markdown2/weasyprint/openpyxl without pinned versions; missing environments will degrade outputs.
- Requires a paid Claude plan and live web search; reachability from mainland China is limited and there is no Chinese-language support.
- Financial figures rely on reference files whose currency cannot be verified statically; cross-check against canada.ca before use.
- Publisher is unverified and this is an individual-maintained project; long-term update commitment is not assured.
What does this skill do, and when should you use it?
This is a Claude skill (financial-planner, v3.0) built specifically for Canadian residents that turns Claude into a personal AI financial coach. It runs a seven-round interview covering income, expenses, debts, assets, insurance, and goals, saving your data as local files so you can pause and resume across sessions. It then produces a complete budget, a phased action plan, an interactive Chart.js HTML dashboard, and shareable PDF/Excel outputs. The skill is deeply Canadian — RRSPs, TFSAs, CPP/OAS, provincial tax brackets and benefits — and enforces double-source verification of any changeable figure. It requires a paid Claude plan and provides financial education, not licensed advice.
At the start of each session it checks the user's folder for 1-my-profile.md to decide whether to resume an interview or enter ongoing coaching, and it politely declines non-Canadian users. It loads 17 reference files on demand (current-figures.md is the single source for all year-specific numbers), runs budget analysis, tax optimization, debt strategy, emergency fund, retirement projections, and insurance gap calculations, and cross-validates results before writing 1-my-profile.md, 2-my-budget.md, 3-my-plan.md, and 4-my-dashboard.html (a self-contained Chart.js dashboard with 15+ sections including net worth projection, CPP timing comparison, and a CRA calendar). It web-searches to verify current ETF, GIC, and HISA data, generates shareable/ PDFs plus a formula-driven Excel budget via openpyxl, and writes an auto-generated README.txt. When processing uploaded documents it extracts only amounts and account types, ignoring SINs and account numbers.
- A young Canadian starting from zero who needs a budget and to decide whether to fund an RRSP or TFSA first
- Someone 50+ weighing whether to take CPP at 60, 65, or 70 and how to avoid the OAS clawback
- A user in financial crisis with credit card or payday loan debt needing triage (negotiation, consumer proposal, or bankruptcy)
- A freelancer or incorporated owner deciding between salary and dividends, handling HST/GST, or claiming SRED credits
- A first-time buyer deciding rent vs. buy, negotiating a mortgage renewal, or weighing prepayment against investing
- Cross-border users: snowbirds wintering in the US, US citizens in Canada facing FBAR/FATCA, or newcomers needing financial onboarding
What are this skill's strengths and limitations?
- Deep Canadian localization: tax brackets, benefits, and health coverage for all 13 provinces and territories, including Quebec's separate system (QPP, QST, QPIP, Revenu Québec)
- All plan files are saved locally on your machine, enabling pause/resume across sessions and shareable PDF/Excel outputs for an advisor or spouse
- Built-in number verification: 150+ official source URLs, with a current-figures.md facts file and double-source web verification for volatile figures
- Ships with 12 test personas and a complete sample family plan (demo-sample-plan/) so you can evaluate output quality before committing
- Clear guardrails: no individual stock recommendations, no tax filing, and proactive referral to a CFP/CPA or fee-only advisor for complex situations
- Strictly Canadian-only; non-residents are declined outright and there is no adaptation for other countries
- The full flow requires a 30–60 minute interview over 2–3 sessions, which may feel heavy for users with quick questions
- Cowork mode requires a paid Claude plan (Pro from about C$24/month); free-tier users cannot use the skill at all
- The dashboard loads Chart.js from a CDN, so shared HTML files may not render charts when opened offline
- It is an education tool, not licensed advice, and cannot access bank accounts or CRA directly
How do you install this skill?
Option 1 (recommended, Cowork mode, Mac/Windows): download financial-planner.skill from the repo, open the Claude desktop app, enter Cowork mode, then Customize → Skills → Upload and select the file. Option 2 (Claude Code CLI): run npm install -g @anthropic-ai/claude-code, sign in with claude, then git clone https://github.com/cjpatten/canadian-finance-planner-skill.git ~/.claude/skills/financial-planner (Windows requires WSL with paths like /mnt/c/Users/YOUR_USERNAME/...). Option 3 (plugin marketplace): run /plugin marketplace add cjpatten/canadian-finance-planner-skill in Claude Code. The free Claude tier does not support skills or Cowork mode — Pro, Max, Team, or Enterprise is required.
How do you use this skill?
Launch Claude from the folder where you want your financial files saved and say "Help me with my finances." First use confirms you're a Canadian resident, then runs a 30–60 minute interview across 1–3 sessions that you can pause anytime — progress is saved locally after every round. Afterwards, return with prompts like "Let's do a monthly check-in," "Should I buy a $900 TV?" (purchase analysis), "Is my RRSP on track?", or "I got a raise" (recalculation). A quick one-off question is fine; the skill won't force a full interview.
How does this skill compare with similar options?
The README positions the skill against human financial planners, noting they typically charge $2,000–$5,000 for comparable planning, and it advises consulting a fee-only advisor for investments over $500K. It deliberately avoids endorsing any platform, saying "managed investing through an online platform" instead of "robo-advisor."