Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/eddmpython/dartlab/companynpx skills add eddmpython/dartlab --skill companygit clone --depth 1 https://github.com/eddmpython/dartlabWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.10801 |
| Opus 5 | $0.00000 | $0.05400 |
| Sonnet 5 | $0.00000 | $0.02160 |
| Haiku 4.5 | $0.00000 | $0.01080 |
Grade A, and why
company scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 805 lines — stays where its author put it; the contents beside it link to each section on GitHub.
엔진 역할
Company는 DartLab의 단일 기업 facade다. 사용자는 종목코드나 ticker로 Company를 만들고, 이 객체에서 원자료 조회, 공시 탐색, 분석, 신용, gather, quant, macro, story를 호출한다.
한국 DART 회사와 미국 EDGAR 회사 모두 같은 facade 패턴을 따른다. 세부 topic/컬럼은 시장과 provider별로 다를 수 있으므로 반환값의 source와 기준일을 확인한다.
공개 호출 방식
import dartlab
c = dartlab.Company("005930")
raw = c.panel("BS")
selected = c.select("매출액")
trace = c.trace("IS")
analysis = c.analysis("financial", "수익성")
credit = c.credit()
price = c.gather("price")
quant = c.quant("모멘텀")
macro = c.macro("사이클") # 시장 매크로 (사이클/위기/시나리오/유동성/심리)
sensitivity = c.analysis("macro", "매크로민감도") # 기업 단위 매크로 민감도는 analysis 엔진
story = c.story()
reportModel = c.reportModel() # 전문 리포트 계약(ReportModel, schemaVersion=2)
industry = c.industry()
강행 호출 룰 (agent 답변 품질 회귀 차단)
단일 종목 질문은 본 엔진이 1 차 진입점. 다음 4 룰 강행:
EngineCall(apiRef="Company.panel", args={"stockCode": "...", "topic": "..."})1 회로 다수 답변 가능 - 응답datadict 에dcrBadge(Track G 7 축 신용) +industryBadge(Track E 산업/lifecycle/peers) 자동 부착. 신용·산업 질문은 추가 EngineCall 불필요.- 본문 안 숫자 / 점수 / 등급 / peers 명에 inline ref 표기 필수 - tool result 의
refs배열에 들어온 id 그대로table:Company.panel:005930또는tableRef:idliteral 로 표기한다. - 다중 종목 비교는
PeerCompareN1 회 강제 (2~12 종목) - Company.panel 를 N 회 반복 호출 + RunPython 정렬 패턴 금지. 메모리 압박 + refs 가치체인 약화. - RunPython 직접 BS/IS/CF 비율 계산 금지 - Company.panel 결과의
dcrBadge.axes(7 축 신용) 또는 Company.analysis 결과의items/history인용. 같은 비율 재계산은 raw fallback 만.
호출 동작
Company 생성 시 target과 market/provider를 확정한다. 이후 panel/select/trace는 원자료 조회, analysis/credit/quant/macro/story는 하위 엔진 호출, gather는 보조 데이터 수집, disclosure/liveFilings/readFiling은 공시 접근을 담당한다.
무인자 또는 topic 누락 호출은 가능한 topic/axis 가이드를 반환할 수 있다. 데이터가 없으면 결손을 0으로 채우지 않고 빈 DataFrame, None, flags, 제한 메시지로 표현한다.
c.panel() 응답 data 의 자동 부착 필드 (단일 종목 답변의 1 차 진입점)
c.panel(topic) 의 반환 dict (server agent/MCP 경유 시 EngineCall(apiRef="Company.panel") 결과의 data) 에는 원자료 외에 다음이 자동 부착된다 - 별도 도구 호출 없이 그대로 인용:
What ships with it
9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 805 lines · 0 tokens per session scan A b390c9bee33d
company is a skill published in the GitHub repository eddmpython/dartlab (209 stars, last pushed 10d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 10,801 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
dividend-analysis
Comprehensive capital allocation analysis: dividends, buybacks, M&A, debt management, and FCF deployment.
position-ladder
Plan a staged position — entry ladder, share-count floor/ceiling, and the trim/re-add cycle that lowers average cost — with lot accounting, wash-sale checks, total-return honesty, and a thesis-break exit gate.
report-generator
Generate professional HTML/PDF investment reports with interactive visualizations.
stock-eval
Evaluate US stocks with comprehensive fundamental and valuation analysis.
stock-valuation
Multi-method stock valuation using DCF, comparable company analysis, EV multiples, and residual income models.
technical-analysis
Technical analysis of US stocks using charts and indicators.