tmdgusya/init-context-skill/init
Claude Code and agent skill by tmdgusya/init-context-skill
- community
Checking that the SKILL.md file is still in the repository.
Use when the user wants to create, update, or review an AGENTS.md context file for their codebase, or when starting a new project that needs agent instructions
Install
npx vskill@latest install tmdgusya/init-context-skill/initSKILL.md
Generate AGENTS.md
Core Principles
- Default is empty. Only include information whose absence causes agent failure.
- Detect, don't generate. Extract facts from config/CI files. Never fabricate content.
- Ask, don't assume. Use AskUserQuestion for what files cannot reveal.
- Format, don't narrate. Bullet points only. No prose, no rationale, no architecture descriptions.
- Differential Context.
Include = (project practice - language/framework default) + prohibitions
Workflow
Phase 1: Scan (automatic)
Detect build/test/lint commands from these sources (priority order):
.github/workflows/*.yml/.gitlab-ci.yml— CI is the source of truthMakefile/justfile— task runner targetspackage.jsonscripts /pyproject.toml[tool.*] /Cargo.toml- Linter config existence:
.eslintrc,ruff.toml,.prettierrc,biome.json
Only extract: command strings, tool names, runtime versions. Do NOT read source files.
Phase 2: Ask (3 questions via AskUserQuestion)
Q1 (required): Present detected commands for confirmation. "I detected these commands. Please confirm or correct:"
- Test:
{detected} - Lint:
{detected} - Build:
{detected}
Q2 (required): "Are there files or directories that must never be modified? (e.g., generated code, vendored deps, contract files)"
Q3 (required): "Are there project practices that deviate from language/framework defaults? Only things agents would likely get wrong. (e.g., 'use Zustand not Redux', 'Result type for errors — no throw', 'Repository pattern only — no direct ORM in handlers'). Answer 'none' if not applicable."
Phase 3: Generate
Combine scan results + user answers. Rules:
- Only
Commandssection is always present (if any commands exist) ConstraintsandConventionssections appear only if user provided content- No content beyond what was detected or user-confirmed
- Target: 100-250 words. Hard limit: 300 words.
Output Template
Anti-patterns
| Never do this | Why | |
|---|---|---|
| List directory structure | Agents discover this via filesystem tools | |
| Describe architecture | Increases exploration, reduces success rate | |
| Add style guidelines | Linters already enforce this | |
| Summarize README/docs | Duplicates existing information | |
| Auto-generate without user confirmation | LLM-generated context is harmful | |
| Exceed 300 words | Verbosity degrades agent performance | |
| Include "nice to have" info | If removal doesn't cause failure, don't include it |
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/init/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/tmdgusya/init-context-skill/init)