Anthropic Open-Sources 10 Wall Street Agents

The Breakthrough

Anthropic restructured the financial-services GitHub repo into ten named end-to-end agents with titles like Pitch Agent, Market Researcher, Earnings Reviewer, Model Builder, Meeting Prep, GL Reconciler, Month-End Closer, Statement Auditor, Valuation Reviewer, and KYC Screener; each agent ships with system prompts, markdown skills bundles, slash commands, and licensed MCP data connectors including FactSet, S&P Global, Morningstar, Moody's, PitchBook, LSEG, Daloopa, Aiera, Chronograph, MT Newswires, and Egnyte.

What Actually Worked

• Pitch Agent runs comps, precedents, and an LBO end-to-end and outputs a branded pitch deck that a banking analyst would build over a weekend.
• Market Researcher takes a sector or theme and produces an industry overview, competitive landscape, peer comps, and a short list of investment ideas.
• Model Builder runs DCF, three-statement, and comps live in Excel and hits XLSX author headlessly when in managed agent mode.
• GL Reconciler finds breaks in the general ledger, traces the root cause, and routes for signoff.
• Every named agent is a self-contained Claude Code plugin with an agent.md file declaring the system prompt, a skills folder with bundled markdown files, a commands folder with slash commands, and an mcp.js declaring allowed connectors; skills sync from seven vertical plugins (investment banking, equity research, private equity, wealth management, fund admin, operations) via sync_agent_skills.py.
• Deployment uses deploy_managed_agent.sh to push to /v1/agents API or check_pui; install requires three commands: add marketplace, install financial-analysis-core bundle with data connectors, cherry-pick agents and verticals.

Context

Wall Street analyst workflows remained a closed shop for twenty years, requiring $25k/year Bloomberg terminals, proprietary Excel templates from bulge bracket banks, Cap IQ subscriptions, and years of training on internal conventions; vendors never published the methodology. Anthropic open-sourced the skill ladder, prompts, conventions, audit checklists, and MCP plumbing for licensed data pulls in an Apache 2.0 repo that hit 11,400 stars and 1,500 forks. Builders in vertical AI for finance, SaaS sales to financial services, middle office, or agentic frameworks can treat the repo as a textbook, though connectors require paid licenses and agents need prompt tuning for production.

Notable Quotes

• "None of this constitutes investment advice and that every output is staged for human signoff."
• "The methodology lived inside the banks what Anthropic just published is the entire skill ladder the workflow itself the prompts and conventions the audit checklist and the plumbing."
• "The trade-offs are real the connectors all require license seats and those subscriptions still cost real money so you cannot clone this repo and walk away with a turnkey terminal stack out of the box."

Content References

See content_references field.