SDD Steering Rules

VerifiedSafe

Rules for creating and managing project steering. What-only principle, truth hierarchy, and Bootstrap/Sync/Rewrite modes.

Sby Skills Guide Bot
DevelopmentIntermediate
7/25/2026
Claude Code
#project-steering#ai-memory#project-management#agent-skills

Recommended for

Automated summary

Manages project memory for AI agents by creating and maintaining steering files that capture project facts, decisions, and patterns without implementation details.

Strengths

  • Ensures AI retains context across sessions
  • Separates what from how to avoid noise
  • Supports multiple modes for new/existing codebases
  • Enforces file size limits to keep steering concise

Limitations

  • Requires initial setup and discipline to maintain
  • May not cover all dynamic project changes
  • Relies on manual sync decisions to avoid automatic drift correction
When to use it

When you want an AI agent to consistently remember project direction without re-analyzing the codebase each session.

When not to use it

For short-lived or rapidly changing projects where maintaining steering files outweighs benefits.

Security analysis

Safe
Quality score90/100

The skill exclusively describes creation and management of markdown files in a project-local .agents directory. It uses Read, Write, Edit, Glob, Grep, and LS - no destructive commands, data exfiltration, or obfuscation. Bash is listed as allowed but no shell commands are present.

No concerns found

How this analysis is made

Examples

Bootstrap New Project
I'm starting a new TypeScript memo app. Please initialize the project steering.
Bootstrap Existing Codebase
Analyze my existing Python FastAPI project and create steering files.
Rewrite with New Intent
I have a new intent: add real-time collaboration to my React dashboard. Update steering.

name: sdd-steering-rules description: 스티어링 생성 및 관리 규칙. What-only 원칙, 진실의 계층(steering→spec→code), Bootstrap/Sync/Rewrite 모드. steering 커맨드에서 사용. allowed-tools: Bash Read Write Edit Glob Grep LS

스티어링 관리 규칙

핵심 원칙

스티어링 = 프로젝트 메모리. 문서화가 아님. AI가 매 세션마다 코드베이스를 처음부터 분석하지 않아도 되도록.

What-Only: 사실, 결정, 패턴만. 구현 방법 없음.

스티어링 저장 위치

스티어링은 AgentSkills 스킬로 관리한다:

.agents/skills/project-steering/
├── SKILL.md              ← 목록 + 동적 로딩 규칙
└── references/
    ├── product.md        (기본 — 요구사항·스펙 관련 작업 시 무조건)
    ├── tech.md           (기본 — 설계·구현 결정 시 무조건)
    ├── structure.md      (기본 — 파일·폴더 생성 시 무조건)
    └── <optional>.md     (선택 — 관련될 때만)

project-steering/SKILL.md 역할: LLM이 언제 어떤 스티어링을 읽을지 결정하는 인덱스.

모드 감지

| 상황 | 모드 | 동작 | |------|------|------| | project-steering/ 없음 + 코드 없음 + --intent 없음 | Bootstrap:New | 사용자에게 프로젝트 목적 질문 후 생성 | | project-steering/ 없음 + 코드 있음 | Bootstrap:Existing | 코드베이스 분석 → 자동 생성 | | project-steering/ 없음 + --intent 있음 | Bootstrap:Intent | intent 기반 즉시 생성 | | 스킬 있음, --intent 없음 | Sync | 드리프트 감지 및 정렬 확인 | | --intent "<새 방향>" 있음 (스킬 있을 때) | Rewrite | 새 방향 중심으로 재작성 |

"코드 없음" 판단: 소스 파일 (.ts, .js, .py, .go 등) 이 존재하지 않는 경우.

Bootstrap:New 모드 — 새 프로젝트

코드가 없고 --intent도 없을 때: 먼저 물어본다.

이 프로젝트에서 무엇을 만들 건가요?
예: "TypeScript로 메모 앱", "Python FastAPI 서버", "React 대시보드"

답변을 받으면 아래 Bootstrap:Intent 방식으로 진행.

Bootstrap:Existing / Bootstrap:Intent 모드 — project-steering 생성

  1. 코드베이스 분석 (있으면) 또는 --intent / 사용자 답변 활용
  2. .agents/skills/project-steering/ 디렉토리 생성
  3. references/ 아래에 기본 스티어링 3개 작성:
    • product.md — 제품 목적, 핵심 기능, 타겟 사용자, 핵심 제약
    • tech.md — 기술 스택, 주요 결정, 외부 서비스
    • structure.md — 폴더 구조, 네이밍 규칙, 모듈 경계
  4. SKILL.md 작성 — 아래 형식 사용:
---
name: project-steering
description: [프로젝트명] 스티어링. 제품/기술/구조 방향. spec 작업 시작 전 로드.
---

# [프로젝트명] Project Steering

## 기본 스티어링 (항상 읽기)
spec 작업 시작 전, 구현 결정 전 무조건 읽는다.

- `references/product.md` — 제품 목적·기능·제약 (요구사항 생성 시)
- `references/tech.md` — 기술 스택·결정·제약 (설계·구현 시)
- `references/structure.md` — 폴더 구조·네이밍·경계 (파일 생성 시)

## 옵셔널 스티어링 (관련될 때만)
현재 작업과 관련된 것만 읽는다.

(없음 — /sdd:steering --add <topic> 으로 추가)

What-Only 수락 기준

✅ 허용 (What)

  • "인증은 JWT, 만료 7일"
  • "데이터베이스는 PostgreSQL 15"
  • "feature-first 폴더 구조"
  • "API 응답은 표준 JSend 형식"

❌ 거부 (How)

  • 코드 블록 (어떤 언어든)
  • 구현 단계 ("먼저 X를 하고...")
  • 함수 시그니처
  • 의사코드

파일 크기 규칙

  • 각 references 파일 ≤100줄
  • 100줄 넘으면 /sdd:steer-trim 실행
  • 핵심만 스티어링에, 상세는 별도 references 파일로

Sync 모드 주의사항

⚠️ 스티어링을 코드에 맞춰 자동 업데이트하지 않음

드리프트 발견 시:

  • 코드 ≠ 스펙 → "코드 수정 필요" 보고
  • 스펙 ≠ 스티어링 → "스펙 또는 스티어링 검토 필요" 보고
  • 의식적 방향 전환이 있었을 때만 스티어링 업데이트

스티어링 참조 방법

다른 스킬에서 스티어링 읽기:

Read `.agents/skills/project-steering/SKILL.md`
→ 기본 스티어링은 무조건 references 파일 읽기
→ 옵셔널은 목록에서 관련성 판단 후 선택적 읽기

개별 스티어링 파일 템플릿

→ references/steering-product-template.md → references/steering-tech-template.md → references/steering-structure-template.md

진실의 계층 및 상세 가이드

→ references/truth-hierarchy.md

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/sdd-steering-rules-mrzxuq0l' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/sdd-steering-rules-mrzxuq0l/download' --output '.skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md.part' && mv '.skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md.part' '.skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

Tools declared by the author (check availability in your agent): Bash Read Write Edit Glob Grep LS

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md' && test -s '.skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/sdd-steering-rules' && cp -R '.skill-downloads/sdd-steering-rules-mrzxuq0l' '.claude/skills/sdd-steering-rules'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/sdd-steering-rules-mrzxuq0l'

openskills read 'sdd-steering-rules-mrzxuq0l'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/sdd-steering-rules-mrzxuq0l/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills