/10k-digest

description: Deep-read a 10-K annual report and output a structured markdown digest — abstract, section summaries, key metrics, and sourced references. Output in English or Traditional Chinese.

10-K Report Digest

⚠️ Data Verification — Do This Before Any Analysis

Before running the digest, confirm the filing details:

1. **Identify the filing** — confirm ticker, fiscal year-end, and exact 10-K filing date from SEC EDGAR. 2. **State your source** — note where the document came from (e.g., "SEC EDGAR, FY2024 10-K filed 2025-02-05") at the top of the output. 3. **Flag incomplete data** — if only partial sections are available, list what is missing before proceeding.

> ⚠️ **Source unavailable.** If the full 10-K cannot be retrieved, display this warning and proceed only with the sections provided.

---

📋 Data & Sources Header — Open Every Output With It

The first thing in the output is this provenance block, filled in — never left as placeholders. It is the standard documented on the [Data & Accuracy](https://yennanliu.github.io/InvestSkill/data-and-accuracy.html) page and the first thing `result-validator` looks for:

Data & Sources
  As of:      <date the figures represent, e.g. 2026-06-30>
  Source:     <primary docs — SEC EDGAR 10-K/10-Q, company IR, FRED, exchange data …>
  Retrieval:  <pasted by user | web/tool retrieval | model memory>
  Confidence: <HIGH | MEDIUM | LOW>

• `Retrieval: model memory` must be paired with `Confidence: LOW` — memory is a placeholder until confirmed against a primary source.
• Mixed sources: list each with its own as-of date rather than blending them.
• Sources disagree on the same figure (a closing price, a deal term): prefer the primary source (SEC filing, company release, exchange data) over an aggregator, then the most recent dated one. Show both values with their sources and dates, say which you used and why, and never average them.
• Data the user pasted is reported as `pasted by user`; do not upgrade its confidence beyond what the user's own source supports.

---

Produce a clean, structured markdown document that distills a company's 10-K annual report into an abstract, per-section summaries, a digest of key metrics, and full source references. Designed for quick comprehension — not trading signals. Output language is selectable: English or Traditional Chinese (繁體中文).

🔎 Retrieving the Filing — the Keyless EDGAR Path

This skill fetches nothing by itself. When the user gives only a ticker, obtain the 10-K in one of three ways and record which one in the `Data & Sources` header:

**A. The host can fetch URLs** → `Retrieval: web/tool retrieval`. Every step is free and needs no API key:

1. **Ticker → CIK** — `https://www.sec.gov/files/company_tickers.json`: find the entry whose `ticker` matches (write `BRK-B` for `BRK.B`) and zero-pad its `cik_str` to 10 digits. 2. **Filings index** — `https://data.sec.gov/submissions/CIK##########.json`: inside `filings.recent` the arrays `form`, `filingDate`, `reportDate`, `accessionNumber` and `primaryDocument` are aligned by index. Take the newest row whose `form` is `10-K` (foreign private issuers file `20-F` instead; a `10-K/A` is an amendment). The top-level `fiscalYearEnd` is `MMDD`. 3. **The document** — `https://www.sec.gov/Archives/edgar/data/<CIK without leading zeros>/<accessionNumber without dashes>/<primaryDocument>`: the complete filing as HTML (inline XBRL). The filing's index page is the same folder plus `<accessionNumber>-index.htm`; cite that URL and the accession number in the References. 4. **Structured statement data (optional)** — `https://data.sec.gov/api/xbrl/companyfacts/CIK##########.json`: every US-GAAP fact the company has tagged, by concept and period. Use it to confirm the figures read from Item 8. 5. **Full-text search (optional)** — `https://www.sec.gov/edgar/search/#/q=%22<phrase>%22&forms=10-K` when a filing has to be located by phrase rather than ticker.

Send a `User-Agent` header that identifies the requester (the SEC requires it, e.g. `Name email@example.com`) and stay under 10 requests per second. Never substitute a summary site for the filing itself.

**B. The host cannot fetch** → `Retrieval: pasted by user`. Ask the user to paste the sections needed — or, if they have the InvestSkill repository, to run

node scripts/fetch-edgar.js <TICKER> --form 10-K        # → data/filings/<TICKER>/*.txt + .json (accession, dates, URL)
node scripts/fetch-fundamentals.js <TICKER>             # → data/fixtures/<TICKER>.md — reconciled statements from the XBRL facts

and paste the result. Both helpers are optional, keyless, and live outside the plugin.

**C. Neither** → show the ⚠️ *Source unavailable* warning, set `Retrieval: model memory` and `Confidence: LOW`, and digest only what the user provided.

---

How to Use

# Analyze by ticker (retrieves the latest 10-K via the EDGAR path above, or asks you to paste it)
/10k-digest AAPL

# Paste 10-K text directly
/10k-digest [paste 10-K sections here]

# Specify fiscal year
/10k-digest MSFT FY2024

# Output in Traditional Chinese
/10k-digest NVDA --lang zh-TW

# Save output to file
/10k-digest GOOGL --output googl-10k-digest.md

---

Output Document Structure

The skill produces a single markdown document with the following sections in order.

---

1. Cover Block

# [Company Name] ([TICKER]) — 10-K Digest
**Fiscal Year:** [FY20XX ending MM/DD/YYYY]
**Filed:** [Filing date]
**Source:** SEC EDGAR — [Direct URL to filing]
**Prepared:** [Today's date]
**Language:** English | 繁體中文

---

2. Abstract

A 3–5 sentence executive summary of the entire 10-K. Cover:

• What the company does (core business in one sentence)
• Year-over-year financial performance headline (revenue, profit, FCF)
• The single most important strategic development disclosed in this filing
• The most material risk disclosed
• Overall filing tone (transparent / cautious / optimistic)

---

3. Section Summari