OpenAccountants
openaccountants/openaccountantsA large open library of tax skills (2,000+ across 250+ jurisdictions) with figures cited to the law, also served through a hosted MCP connector.
With caveats
Broad access Depth 2/5
Claude for Financial Services Claude skills
Anthropic's reference agents, skills and data connectors for banking, research, fund admin and KYC, usable as Claude plugins or Managed Agents.
Our verdict Recommended with caveats
Credible, Apache-2.0 reference material from Anthropic, but it is draft-for-review tooling with duplicated skills and connectors that depend on paid data providers.
What Claude for Financial Services does
This repo packages finance workflows two ways from one source: as Claude plugins (Claude Cowork or Claude Code) and as Claude Managed Agent templates you deploy through the API behind your own workflow engine. Ten named agents each bundle the skills they need: Pitch Agent (comps, precedents and LBO into a deck), Meeting Prep, Market Researcher, Earnings Reviewer (call and filings into a model update and note draft), Model Builder (DCF, LBO, three-statement and comps models in Excel), Valuation Reviewer, GL Reconciler, Month-End Closer, Statement Auditor and KYC Screener. Separate vertical plugins hold the underlying skills and slash commands for investment banking, equity research, private equity and wealth management, and partner-built plugins cover LSEG and S&P Global.
The file metrics count 112 SKILL.md files averaging about 8 KB, but many are copies shared across agents (for example audit-xls appears in several plugins), and only a few skills carry reference files. A few skills include Python requirements for Excel handling (openpyxl, requests), and a managed-agent-cookbooks folder has deployment scripts and per-agent security notes. Data connectors point at MCP servers from providers such as Moody's, Morningstar, S&P Global, FactSet and LSEG, which may need a subscription or API key. The README states that nothing here is investment, legal, tax or accounting advice and every output is staged for human sign-off.
Good for
- Finance teams drafting models, memos and reconciliations for human review
- Developers building Managed Agent workflows for banking, research or fund administration
- Learning how a vendor structures finance skills, commands and connectors
Not for
- Automated trading or any decision made without review
- Individuals who need finance help without data subscriptions or an Anthropic API setup
What's inside depth 3/5
112 SKILL.md files averaging 8 KB with some references, Excel-handling Python requirements and Managed Agent cookbooks, but heavy duplication across plugins and no evals.
- Skills
- 112
- Reference docs
- 48
- Scripts
- 8
- Tests / evals
- 0
- Templates
- 2
- Avg SKILL.md
- 8.2 KB
Counted from the repository file tree. How we score depth.
Dependencies and costs
- Claude Cowork or Claude Code, or an Anthropic API key for Managed Agents
- Optional MCP data subscriptions (for example Moody's, Morningstar, S&P Global, FactSet, LSEG)
- Excel and Python (openpyxl, requests) for model-building skills
Security review
Review before installingMarkdown finance skills and agents (DCF and 3-statement models, earnings review, GL reconciliation, KYC screening, pitch decks) plus two small local Python validators. The files we have contain no allowed-tools, hooks or fetch-and-follow instructions. The deployment scripts, Microsoft 365 admin scripts and connector configs referenced by the README were not downloaded, so those parts are unreviewed.
Worth knowing
- Relies on a CLI or package we could not reviewThe Managed Agents path runs scripts/deploy-managed-agent.sh with your ANTHROPIC_API_KEY; the commands and agents also wire in partner MCP data connectors (for example Moody's, LSEG, Morningstar, S&P Global, FactSet); none of that code is in our files.README.md:81
- Makes network requestsConnector endpoints such as api.moodys.com and api.analytics.lseg.com are listed as MCP servers for paid data subscriptions; the skills themselves also fetch SEC filings.README.md:127
What we checked
The repository (363 files) on Oct 29, 2026. Not a full review: 302 of 363 files were not downloaded (large repository; we reviewed the skill folders); we read 0 of 3 code files; it relies on code outside the repository: scripts/deploy-managed-agent.sh, managed-agent-cookbooks/ and vertical-plugins/*/.mcp.json (302 of 363 files were not downloaded); claude-for-msft-365-install admin scripts (scripts/*.sh and *.ps1), described as provisioning the Claude Microsoft 365 add-in and able to run destructive --apply paths on Office data; partner MCP connector services.
NVIDIA SkillSpector static scan: 33 pattern hits (analysis-evasion 14, Excessive Agency 9, Supply Chain 8, Data Exfiltration 2). The 33 hits are low-signal: empty 'analysis-evasion' matches on markdown files, 'without asking' inside an audit skill that says not to change anything without asking, partner API URLs in the README, and unpinned openpyxl/requests in a requirements file.
A review, not a certification. Skills change after we check them — read the files before installing.
License
Apache-2.0 — commercial use allowed. Partner data connectors have their own terms.
Can it run unattended?
Not really. The agents are designed to stage drafts for human sign-off; they do not execute transactions or post to ledgers, and no unattended verification loop is documented in the skills.
Alternatives to Claude for Financial Services
Edgar Change Interpreter
cmdrvl/edgar-change-interpreterA short skill that makes Claude compare SEC 10-K, 10-Q and 8-K filings for material changes and silent risks, with quoted evidence and fixed output sections.
With caveats
Check first Depth 2/5
Wind AliceMarket
Wind-Alice/AliceMarketChinese-language finance skills for A-share and global equity analysis, plus Wind data and agent entry points that need a Wind API key.
With caveats
Check first Depth 2/5
Claude for Legal
anthropics/claude-for-legalAnthropic's suite of legal-workflow plugins: 151 skills across in-house practice areas, each driven by a cold-start interview and a practice profile.
With caveats
Check first Depth 2/5
FAQ
What is Anthropic's Claude for Financial Services repo?
It is a set of reference agents, skills and data connectors for investment banking, equity research, private equity, wealth management, fund administration and KYC. You can install them as Claude plugins or deploy them as Claude Managed Agents.
How do I install the financial services plugins?
In Claude Code, run claude plugin marketplace add anthropics/financial-services, install financial-analysis@claude-for-financial-services first, then add agents such as pitch-agent or gl-reconciler. In Cowork, paste the repo URL under Settings, Plugins.
Does it give investment advice?
No. The README says nothing in the repo is investment, legal, tax or accounting advice, and the agents draft work for review by a qualified professional rather than making recommendations or executing transactions.
Are the financial data connectors free?
No guarantee. The connectors point to MCP servers from providers like Morningstar, S&P Global, FactSet and LSEG, and the README notes access may require a subscription or API key.
Sources: GitHub repository (stars, license, file tree, last push Oct 28, 2026). Review based on the README and SKILL.md files as published.