story

A report builder that combines financial, credit, industry, market, quantitative, and comparison results into a six-part company report. It arranges findings from other analysis tools rather than calculating its own numbers.

In plain words
What is it for?
Use it to create full company reports, executive briefs, valuation or credit reports, individual sections, or custom Markdown and HTML reports.
Why use it?
It gives different types of analysis one consistent report structure and keeps each conclusion linked to its source.

Skill for Claude CodeCodex

Install

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.

agentmods
npx agentmods add skills/eddmpython/dartlab/story
Any agent
npx skills add eddmpython/dartlab --skill story
Clone the repo
git clone --depth 1 https://github.com/eddmpython/dartlab

Made for: Claude Code, Codex.

Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,074 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.04074
Opus 5 $0.00000 $0.02037
Sonnet 5 $0.00000 $0.00815
Haiku 4.5 $0.00000 $0.00407

Measured 2d ago against content hash e4e22311cc8d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

story 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.

src/dartlab/skills/specs/engines/story/SKILL.md · 306 lines

How it starts

The opening of the file, as written. The whole thing — 306 lines — stays where its author put it; the contents beside it link to each section on GitHub.

엔진 역할

story해석하지 않고 다양한 관점의 근거를 배치 하는 L3 조합기다. 분석엔진 X. L2 5 분석엔진 - analysis (재무 인과) · credit (신용 위험) · macro (시장 환경) · quant (정량 신호) · industry (밸류체인 위치) - 와 L1.5 scan (횡단 비교) 결과를 블록 단위로 조합해 6 막 구조 보고서를 만든다.

story 가 L3 조합기인 이유 - 순환참조 방지

L2 5 분석엔진은 도메인 격리 (analysis ⊥ credit ⊥ macro ⊥ quant ⊥ industry) 가 import 단방향으로 강제돼 있다. L2 끼리 직접 import 하면 순환참조가 생긴다. 이 결합 책임을 story 가 단독으로 짊어져 6 막 보고서로 직조한다 - 자체 숫자 계산 0, 모든 숫자는 하위 엔진 결과 ref 에 묶이고, story 자체는 thesis · evidence · risk · limit 의 문장 골격만 제공.

이 구조 덕분에 새 L2 분석엔진을 추가해도 기존 L2 엔진은 수정 0 - story 의 블록 등록만 늘어난다.

공개 호출 방식

import dartlab

# 1. Company-bound - 가장 단순한 진입점
c = dartlab.Company("005930")
story = c.story()                          # 자동 reportType + 기업유형 감지
print(story.render("markdown"))

# 2. reportType 명시
credit_report = c.story(reportType="credit")
valuation = c.story(reportType="valuation")
exec_brief = c.story(reportType="executive")

# 3. 단일 섹션
revenue = c.story("수익구조")               # 수익구조 블록만

# 4. 자유 조립 (블록 단위)
from dartlab.story import blocks, Story
b = blocks(c)
custom = Story([b["growth"], b["marginTrend"], b["cashFlowOverview"]])

# 5. 출력 형식 변환
print(custom.toMarkdown())
print(custom.toHtml())
print(custom.render("rich"))               # 터미널 색상
print(custom.render("json"))               # AI 소비용

강행 호출 룰 (agent 답변 품질 회귀 차단)

story 는 L3 조합기 - 자체 계산 0. 모든 숫자는 하위 엔진 (analysis/credit/macro/quant/industry/scan) ref 인용으로 들어와야 함.

  1. story 본문 안 모든 숫자에 하위 엔진 ref inline 표기 필수 - [tableRef:...]/[valueRef:...] 형식. ref 없는 숫자는 story 본문 진입 차단.
  2. story 안에서 직접 계산 금지 - RunPython 으로 ratio/forecast/score 산출 금지. 하위 엔진 호출 결과의 items/flags/assumptions 그대로 차용.
  3. 블록 evidence 부족 시 빈 섹션 + limits 에 명시 - 임의로 채우거나 환각 금지. story 의 spec 가 "evidence 비면 빈 섹션" 정공.
  4. reportType 미명시 시 자동 감지 결과 본문에 노출 - "자동 선택: executive (이유: ...)" 한 줄.
  5. 같은 (stockCode, period) 분기 데이터는 1 회 fetch 후 in-memory 재사용 - 같은 회사 같은 기간의 EngineCall 을 axis 만 바꿔 반복 호출 금지. story 가 L3 조합기로서 차입금·충당부채·부문정보 등 여러 블록을 만들 때, Company.panel 한 번으로 모든 stmt (BS/IS/CF) + 분기 시계열을 받아 in-memory 분기 후 각 블록에 분배한다. 회귀 사례 - Q3 hybrid 답변 137s + RSS 7.6GB 폭증 (같은 005930 분기 데이터 차입금/충당부채/부문정보 4 회 반복 load). operation.performanceProfile 메모리 안전 강행 (Company 1 개 ≈ 200~500MB) 직결.
  6. 렌즈 수집기는 대형 내부 결과를 보존하지 않는다 - lensProducts 원문과 신용 패널에 필요한 최소 필드만 Company 세션에 남긴다. analysis, industry, quant, macro의 원본 결과 전체를 _lensBundle 또는 별도 Story 캐시에 붙잡아 두지 않는다.

Read the full file on GitHub · 306 lines

Files

What ships with it

2 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.

Changes

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.

  1. 2d ago First seen · 306 lines · 0 tokens per session scan A e4e22311cc8d

Subscribe to this mod's changes

story 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 4,074 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.