AI analysis grounded in the code graph — computed facts, not vibes · 2026-09-23T03:03:03Z
This is Anthropic's reference implementation of financial-services workflows for Claude: a bundle of named agents (Pitch Agent, Market Researcher, GL Reconciler, KYC Screener, and others) plus the underlying "vertical plugins" — skills, slash commands (/comps, /dcf, /earnings), and data connectors — organised by FSI vertical (investment banking, equity research, private equity, wealth management). Each agent can be installed as a Claude Cowork plugin or deployed via the Claude Managed Agents API (/v1/agents), from the same system prompt and skill set. It's aimed at analysts and firms wanting Claude-drafted models, memos, and reconciliations that still require human sign-off.
The 260-star gain in a single day tracks with active, visible feature churn rather than a single marketing event: commit titles show a "Launch of Claude for Financial Advisors" followed shortly by its removal, alongside new manifest capabilities (available_models config key, disabled_features web_search slug). Given no release notes were captured, the growth is more plausibly attributable to Anthropic's own reach and the repo's positioning as an official reference for a large vertical (finance) than to any documented feature that clearly explains a 35k+ star base — the evidence available doesn't fully account for the scale.
What changed recently, how it's actually built (from the code graph), and whether you should care. Free account — no card, no spam.