Spreadsheets to Close: Claude Code Plugins for Finance

Start Here

Finance work inside Claude Code breaks into a handful of concrete jobs — querying and reconciling numbers, editing spreadsheets, automating month-end close and reporting, and putting guardrails on automated trading logic — and the Claude Code plugins for finance in the directory today fit those jobs unevenly. One queries a data warehouse over an MCP server; another only manipulates Excel files on disk; two more are sprawling agent teams where finance is a single lane among dozens. By the end of this post you should be able to match a plugin to a specific finance task instead of installing a generalist and hoping it covers your close checklist.

Set expectations first: this is an early, thin category. The dedicated finance plugin shows 5 installs in the last 7 days, and the broad multi-agent teams each show 1. None of these has the traction of the larger workflow plugins elsewhere in the directory. So treat the install numbers as signal about maturity, not proof of quality — several of these are capable on paper but barely exercised in public. The eight plugins below are grouped by what their components actually let them do.

The Standouts

The most finance-specific plugin is finance. Its summary describes automating journal entries, account reconciliations, financial statement generation, variance analysis, and SOX compliance testing, and it ships skills plus an MCP server with integrations to BigQuery, Slack, and Benchling. Reach for it when your numbers live in a warehouse and you want Claude to query and reconcile them rather than read a static file. That MCP surface is what separates it from everything else here — it is the only finance-named plugin in this set that talks to an external data source instead of operating purely on local files.

Where finance queries a warehouse, xlsx works the file. It is a skills-only plugin that reads, edits, creates, and converts Excel, CSV, and TSV workbooks with full support for formulas, formatting, and charts. Use it for the unglamorous middle of finance work — cleaning an export, rebuilding a model, or generating a formatted workbook to hand off. It complements finance directly: pull and reconcile with finance over MCP, then shape the output into a deliverable with xlsx, no warehouse credentials required.

For the controller's calendar, fund-admin is the narrow specialist. Also skills-only, it targets fund administration and month-end close: reconciliation break analysis, accrual schedules, roll-forward tie-outs, and variance commentary, with an eye toward review and audit documentation. When your problem is the recurring close rather than ad-hoc analysis, this is the more focused choice than the general-purpose finance plugin, because its skills are written around close-cycle artifacts rather than broad accounting.

The outlier on adoption is product-manager-skills, with 4,405 stars — by far the most-starred plugin in this group. Be honest about fit, though: it is a commands-and-skills product-management library (discovery, PRD creation, roadmap planning, prioritization), not an accounting tool. It earns a place here only for fintech builders who need finance-adjacent product framing, and it is worth contrasting precisely because high stars do not make it a finance plugin — its component coverage is about shipping product, not closing books.

Finally, skill-creator is the meta-pick. It ships agents and skills to author, evaluate, and iteratively improve your own Claude Code skills, with automated blind comparisons and grading against expectations. If none of the prebuilt finance skills match your firm's chart of accounts or reporting format, this is how you build and stress-test a bespoke one. It complements every other plugin here: the others give you finance behavior, while skill-creator gives you a way to manufacture more of it. For background on what a skill versus an agent versus a command actually buys you, the plugin examples by component type post is a useful companion.

Trade-offs to Know

The clearest trade-off is integration depth versus self-containment. The finance plugin's skills-plus-MCP design lets it reach BigQuery, Slack, and Benchling, which is powerful but means setup, credentials, and a connected warehouse before it does anything. xlsx, being skills-only, runs against a file you already have — lower ceiling, near-zero setup. If your financial analysis lives in a spreadsheet today, the local plugin gets you moving faster; if it lives in a warehouse, the MCP plugin is the one worth the configuration. The broader directory has 9,419 MCP components against 242,546 skills, so this MCP-versus-skills split is a directory-wide pattern, not unique to finance.

The second trade-off pits two multi-agent teams against each other. enterprise-team deploys 75 specialized agents across engineering, product, infrastructure, security, finance, legal, and HR, auto-routing tasks to domain experts; it carries 17 stars and 1 install in the last 7 days. tonone orchestrates a comparable full-spectrum team of more than 100 agents and adds hooks alongside its agents and skills, also at 1 install in the last 7 days. The grounded difference: tonone's hooks let it wire into lifecycle events, while enterprise-team explicitly enumerates finance as one of its routed domains. For either, finance is a slice of a much larger surface — closer to hiring a whole department than a finance tool. Both lean heavily on agents, so the agents directory is the place to judge whether that breadth fits your workflow, and the automate Claude Code workflows post covers how this kind of orchestration plays out in practice.

The third is about maturity. trading-agent-safety is the only entry aimed squarely at Claude Code trading tools, framed as safety guardrails for automated trading logic. It is skills-only and sits at 0 stars — an unproven, single-purpose plugin. The honest read is that it names a real and underserved gap (risk controls around automated trading) but has no adoption signal behind it yet, so vet its skills before trusting them with anything live.

Which Claude Code Plugins for Finance to Reach For

Match the plugin to the component type that fits your task:

• You have a spreadsheet to clean or a model to rebuild. Reach for the skills-only xlsx — it edits workbooks, formulas, and charts directly with no external setup.
• Your numbers live in a warehouse and need querying or reconciling. The MCP-backed finance plugin is the only one here that connects to BigQuery and Slack to pull and reconcile live data.
• You run a recurring month-end close. fund-admin is built around close-cycle artifacts — break analysis, accruals, roll-forward tie-outs, and audit-ready commentary.
• You need a finance skill that does not exist yet. Use skill-creator to author and grade a custom one against your own expectations.
• You want a whole simulated department, finance included. enterprise-team or tonone route finance work alongside engineering, legal, and security — broad rather than deep.
• You are building automated trading and want guardrails. trading-agent-safety is the only on-topic option, but it is unproven at 0 stars — review before relying on it.

Wrap-up

Finance is a young corner of a directory that tracks 39,259 plugins, and it shows: install counts here are in the single digits, and a couple of the entries are general-purpose agent teams or a product-management library that merely touch finance. The dependable picks are the focused, single-job skills — xlsx for spreadsheets, fund-admin for the close — plus the MCP-connected finance plugin when your data lives outside local files. The trading-safety angle is real but unvalidated, and skill-creator is the escape hatch when nothing prebuilt fits. Pick by the job in front of you and the component type that serves it, browse the rest of the directory for adjacent tools, and re-check the install numbers as this category matures.