agentleFS
ray5273ray5273/kr-research-kit/skills/kr-foreign-analyst/SKILL.md

kr-foreign-analyst

Skill2 starsChanged 8 days agoMITView on GitHub

Collect foreign-IB research coverage (Morgan Stanley, Goldman, JPMorgan, Nomura, CLSA, UBS, HSBC, Macquarie, Citi, BofA, Daiwa, etc.) on KRX-listed companies from Korean news (연합인포맥스/한경/머니투데이/네이버 뉴스) and render it as a Markdown block that drops directly into the `## Street / Alternative Views` section of a kr-stock-analysis memo. Use ad-hoc per company while writing or updating a memo. Requires kr-naver-browse. Never invent a broker, rating, or target price — every view must link back to a dated Korean news article that named the broker.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not use this skill to:

    line 29
  2. Distinctive:

    Never invent a broker, rating, or target price: Every record must trace back to a Korean news article that named the broker.

    line 70
  3. Distinctive:

    Every bullet cites a dated news URL: If no date was parseable, fall back to the article's publication date, not today.

    line 72

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Korean Foreign Analyst Coverage

Given a company name (and ideally its KRX ticker), find articles in Korean media that quote or summarize foreign investment-bank research on the stock, extract structured broker metadata (name, rating, target price, date, thesis snippet), and produce a Markdown block that pastes directly into ## Street / Alternative Views.

When To Use

  • You're writing or updating a kr-stock-analysis memo and want a fast audit of what foreign IBs are publishing on the name.
  • You want to refresh the existing Street / Alternative Views block with the latest broker takes surfaced by Korean media.

Do not use this skill to:

  • Produce a final thesis — synthesis belongs in the memo.
  • Cite foreign IB reports as primary sources for numbers. The skill only reads what Korean outlets chose to quote; the actual research is upstream and usually behind a paywall.
  • Blanket-monitor many stocks at once. Call it ad-hoc per company. For portfolio-wide monitoring, use kr-portfolio-monitor.

Pipeline

  1. Fetch — scripts/fetch-analyst-coverage.js runs up to four Naver News queries against the company name (<name> 외국계, <name> 외국계 리포트, <name> 목표주가, OR-expansion over the IB whitelist), de-duplicates articles, fetches each body through kr-naver-browse, and extracts broker mentions. For each broker found, it records { broker, rating, targetPriceKrw, reportDate, articleDate, articleUrl, press, title, snippet, sourceQueries }. Articles are URL-keyed in a global cache; discovery results are per-ticker per-day.
  2. Summarize — scripts/summarize-analyst-views.js reads the coverage JSON and renders a Markdown block with one `Street view` bullet per broker view plus one `Bottom line` bullet synthesizing consensus / disagreement.

Output Contract

The summarize step writes a Markdown file that looks like:

## Street / Alternative Views

- `Street view`: Morgan Stanley (2026-04-15, Overweight, TP 95,000 KRW): 모건스탠리는 AI 수요와 마진 회복을 근거로 비중확대 의견을 유지한다. — 연합뉴스, 2026-04-16, https://n.news.naver.com/mnews/article/001/0012345678
- `Street view`: Goldman Sachs (2026-04-12, Buy, TP 102,000 KRW): HBM 점유율 상승과 파운드리 수주 회복을 근거로 매수 의견. — 한국경제, 2026-04-12, https://www.hankyung.com/...
- `Bottom line`: 외국계 뷰는 갈림 — Morgan Stanley는 Overweight; Goldman Sachs는 Buy; JPMorgan는 Neutral. TP 중앙값 95,000 KRW. Filing 확정 범위는 별도 확인 필요.

Reference: references/output-format.md, references/ib-aliases.md.

Operating Rules

  1. Never invent a broker, rating, or target price. Every record must trace back to a Korean news article that named the broker.
  2. Every bullet cites a dated news URL. If no date was parseable, fall back to the article's publication date, not today.
  3. Deduplicate by (broker, reportDate, targetPriceKrw). Merge duplicates by keeping the longest snippet, earliest article date, and union of source queries.
  4. Scope rating/TP extraction to the paragraph around the broker mention. A single article may quote multiple IBs with different targets — do not cross-contaminate.
  5. Truncate snippets to 240 characters in JSON and 240 in Markdown. Quote verbatim; do not paraphrase.
  6. Respect the cache. Same article URL is never re-fetched without --no-cache.
  7. Fail loud on empty coverage by default. The harness all flow passes --allow-empty so small caps with genuinely no foreign-IB coverage do not block memo generation.
  8. Sort records by reportDate descending, broker name ascending as tie-break.

Invocation

# Step 1: fetch (standalone)
node skills/kr-foreign-analyst/scripts/fetch-analyst-coverage.js \
  --company "삼성전자" --ticker 005930 --output /tmp/fa.json --verbose

# Small cap, tolerate empty result
node skills/kr-foreign-analyst/scripts/fetch-analyst-coverage.js \
  --company "한미글로벌" --ticker 053690 --output /tmp/fa.json --allow-empty

# Step 2: summarize
node skills/kr-foreign-analyst/scripts/summarize-analyst-views.js \
  --input /tmp/fa.json --output /tmp/fa.md

# Via harness (chains both)
node scripts/harness.js --mode foreign --ticker 005930 --company "삼성전자"

# As part of full memo pipeline
node scripts/harness.js --mode all --ticker 005930 --company "삼성전자" --with-foreign

Fail-Loud Rules

  • --company missing → exit 1.
  • --output missing → exit 1.
  • gstack browse binary missing → exit 1 (surfaced from kr-naver-browse).
  • Zero coverage records → exit 1 unless --allow-empty.
  • Malformed --input JSON (summarize step) → exit 1.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.