tidy · davepoon — Claude Code skill for transaction cleanup
This Claude Code skill groups uncategorized transactions and suggests categories. It starts when asked to tidy transactions or sort uncategorized activity.
MIT 104 installs ★ 3,785 updated 2026-10-25
npx skills add davepoon/buildwithclaude --skill tidy -a claude-codeWhat tidy does
tidy finds transactions without categories, groups them by similar descriptions or merchant names, and suggests labels from the user’s existing categories. It shows each group’s transaction count and total, then asks whether to approve, change, or skip the suggestion.
For recurring activity, the skill can propose a rule. It previews how many existing transactions would match, and creates the rule only after approval. One-off transactions can be categorized directly, with a party name added if approved. Updates use the MCP query and admin tools; web and email searches can help identify unclear transactions when available.
Afterward, tidy reports the number of categorized transactions, rules created, and uncategorized transactions left. User approval is required before changes are made.
What it helps with
- group similar transactions that lack categories
- suggest categories for transactions by merchant
- create rules for recurring payments
- assign a party name to a one-off transaction
Example prompts for the agent
- Tidy up my uncategorized transactions for this month.
- Group transactions without categories and suggest a category for each group.
- Review uncategorized expenses and show which groups could use rules.
How to install tidy
1Skills CLI (Vercel)
npx skills add davepoon/buildwithclaude --skill tidy -a claude-code2Manually: put the files in place
~/.claude/skills/tidy/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/tidy/
1Skills CLI (Vercel)
npx skills add davepoon/buildwithclaude --skill tidy -a codex2Manually: put the files in place
.agents/skills/tidy/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add davepoon/buildwithclaude --skill tidy -a cursor2Manually: put the files in place
.cursor/skills/tidy/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install davepoon/buildwithclaude plugins/cashflow/skills/tidy2Manually: put the files in place
.github/skills/tidy/SKILL.md1Skills CLI (Vercel)
npx skills add davepoon/buildwithclaude --skill tidy -a gemini-cli2Platform CLI
gemini skills install https://github.com/davepoon/buildwithclaude --path plugins/cashflow/skills/tidyManually: put the files in place
.devin/skills/tidy/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add davepoon/buildwithclaude --skill tidy -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about tidy
How do I install tidy in Claude Code?
Run npx skills add davepoon/buildwithclaude --skill tidy -a claude-code. The skill file is at ~/.claude/skills/tidy/SKILL.md.
What does tidy do?
It groups similar uncategorized transactions, suggests categories, and can propose rules for recurring activity. It asks for approval before making changes.
What does tidy need to work?
Transaction handling uses the MCP query and admin tools. Web or email search can help identify unclear transactions when those tools are available.
Source text
SKILL.md · MIT
---
name: tidy
description: Triggered by "tidy up", "clean up transactions", "categorize uncategorized", "organize my transactions"
---
# Tidy Up Uncategorized Transactions
Batch-categorize uncategorized transactions by clustering similar ones and applying categories in bulk.
## Workflow
1. **Fetch uncategorized transactions.** Call the `query` MCP tool:
```json
{ "detail": true, "is_uncategorized": true, "period": "last_90d", "limit": 200, "sort": "-amount" }
```
If `$ARGUMENTS` contains a time period (e.g. "this month", "last 30 days"), use that instead of `last_90d`.
2. **Research unknown transactions.** For transactions you can't identify from the description alone:
- **Web search first** (if available): Search for the merchant name, any phone numbers or domains in the description, or the raw description itself. This often reveals the business behind cryptic processor names.
- **Search the user's email** (if available): Search for the party/merchant name to find order confirmations or receipts. If that doesn't match, search for the exact dollar amount (e.g. "$47.23") to find receipts that way.
3. **Cluster by pattern.** Group the results by normalized description or party name. For each cluster, note the count and total amount.
4. **Suggest categorization.** For each cluster, propose:
- A **category** (pick from the user's existing categories)
- A **party** name (the clean merchant/counterparty name)
5. **Present to the user.** Show a table or list of clusters with:
- Pattern / merchant name
- Count of transactions
- Total amount
- Suggested category
- Whether you recommend creating a rule
Ask the user to approve, modify, or skip each cluster.
6. **Prefer rules over one-off annotations.** If a cluster has more than one transaction, or the merchant is likely to appear again (subscriptions, regular stores, utilities, etc.), create a rule rather than annotating individual transactions. Rules automatically categorize future transactions too.
- Preview first: `admin { "entity": "rule", "action": "preview", ... }`
- Show the preview (how many existing transactions would match)
- If user confirms, create: `admin { "entity": "rule", "action": "create", ... }`
7. **Annotate the rest.** For truly one-off transactions where a rule wouldn't help, apply directly:
```json
{ "action": "categorize", "filter": { "search": "<pattern>" }, "category_name": "<approved_category>" }
```
Also set the party if one was approved:
```json
{ "action": "set_party", "filter": { "search": "<pattern>" }, "party_name": "<approved_party>" }
```
8. **Summarize.** Report how many transactions were categorized, how many rules were created, and how many uncategorized transactions remain.
## Tone
Stick to the facts. Present findings and suggestions without judgement — no commentary on spending habits. Just clear, plain-language observations and actionable options.
© davepoon · MIT · davepoon/buildwithclaude
Similar skills
All skills for Finance and business →tidy by other authors comes first.