NaCl63

Equity Research Report Claude Skill — AI skill for Claude Code

AI community

Claude Agent Skill that turns a ticker into an analyst-grade equity research PDF plus a live Excel model — SEC EDGAR XBRL as the primary source, single-driver valuation modeling, and two blocking veri.

How to install Equity Research Report Claude Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open NaCl63/equity-research-report-Claude-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Equity Research Report Claude Skill does

Alternatives in AI

Graymatter — 30 sec to give your AI agents persistent memory 460 ★
Huashu Excel — 数据分析与 Excel 全流程 skill：体检脏表、清洗、对齐需求、分析、对账、交付 303 ★
Visual Skills — AI film director skills for agents: cinematic dramaturgy (Murch, blocking, montage) + exact prompt syntax for 212 ★

README

equity-research-report

A Claude Agent Skill that turns a ticker into two deliverables: a multi-page equity research PDF and a companion Excel model with live formulas. Financials come from SEC EDGAR's XBRL API rather than scraped summaries, and the workbook has to pass two blocking verification gates before it ships.

The target output is what a sell-side or buy-side analyst would hand you, not a formatted summary of the first page of search results.

What it produces

**PDF report** — nine sections built with `reportlab.platypus`: cover and executive summary, business model mechanics, industry/competitive/regulatory landscape, financial analysis, valuation, technical analysis, catalyst calendar, key risks, and a sources-and-limitations section. Vector line, bar and scenario charts are generated inline, no image embedding. Pages are rendered to PNG and visually checked before delivery.

**Excel model** — native openpyxl charts that reference cells and update when inputs change, heat-mapped sensitivity grids, frozen panes on wide tabs, and depending on which template fits: a single-driver grid, explicit multi-year Bear/Base/Bull projections with two independent valuation methods, or a full 3-statement model with DCF, unlevered FCF, an SBC/dilution schedule and a reverse DCF.

How it works

**Step 1 — Research.** Four parallel research passes: fundamentals and financials, technical and market data, industry/competitive/regulatory landscape, and comps and valuation multiples. EDGAR XBRL is queried first for anything the company has actually filed; the `frames` API pulls peer multiples measured over an identical period, which is what hand-built comps tables usually get wrong. Every material number carries a source URL. Conflicting figures are reported side by side instead of silently resolved, and gaps are stated rather than filled in.

Before anything gets built, the skill identifies the single variable that drives the company's valuation — net interest margin