init-agent-config
Scaffold a new project with the relevant subset of agent-harness rules, hooks, skills, recommendations, and templates.
Skill by jajupmochi
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/jajupmochi-init-agent-config
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/init-agent-config
What's inside
- Skill name
init-agent-config- Description
- Scaffold a new project with the relevant subset of agent-harness rules, hooks, skills, recommendations, and templates. Asks about project type, language preferences, and context tags, then composes a project-specific CLAUDE.md, .claude/settings.json, and starter files. Use in a fresh or existing project to apply agent-harness conventions.
- License
- MIT
- Instructions
- 253 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# /init-agent-config Compose `agent-harness` into a project — interactive scaffold that asks about project type and selects the right subset. ## Master TOC - [Pre-flight](#pre-flight) - [Step 1: Detect the current project state](#step-1-detect-the-current-project-state) - [Step 2: Ask the user 6 questions](#step-2-ask-the-user-6-questions) - [Step 3: Resolve consumption mode](#step-3-resolve-consumption-mode) - [Step 4: Compose the project](#step-4-compose-the-project) - [Step 5: Post-setup verification](#step-5-post-setup-verification) - [Step 6: Show summary + next steps](#step-6-show-summary--next-steps) - [Idempotency](#idempotency) ## Pre-flight Verify (per `plugin-preflight` rule): - `agent-harness` is locally cloned at `~/.claude/agent-harness/` OR the user has internet access for raw-URL imports OR the plugin is installed - The current directory is writable and the user is OK adding `CLAUDE.md` / `.claude/` / etc. If `agent-harness` is not yet cloned and the user wants the local-clone consumption mode: ```bash git clone https://github.com/jajupmochi/agent-harness.git ~/.claude/agent-harness ``` Ask the user before running this. ## Step 1: Detect the current project state ```bash # Check if directory is empty ls -A | wc -l # Check if it's a git repo git rev-parse --is-inside-work-tree 2>/dev/null # Check for existing CLAUDE.md ls CLAUDE.md 2>/dev/null # Detect language / framework hints ls package.json pyproject.toml Cargo.toml go.mod 2>/dev/null ``` Branches: - **Empty directory** → fresh scaffold from a `agent-harness` template - **Existing project, no CLAUDE.md** → add `CLAUDE.md` + `.claude/` on top of existing code - **Existing project, with CLAUDE.md** → ask the user: merge with existing? back up first? skip CLAUDE.md and only do `.claude/`? ## Step 2: Ask the user 6 questions Use `AskUserQuestion` to gather: 1. **Project type** (single-select): - Python research package - Static personal/portfolio site (HTML/CSS/JS) - Frontend application (React / Vue / etc.) - Other / custom (skip template, just add rules + tooling) 2. **Bilingual policy** (single-select): - Bilingual (EN canonical + zh mirror) for top-level docs - English only - Chinese only - Other / let me decide later 3. **Final-output language preference** (single-select): - Chinese (中文) — apply `chinese-output` rule - English — skip the rule - Other / let me decide later 4. **Context tags** (multi-select — only the matching items get installed): - `always` (universal rules: pre-edit-confirmation, phased-planning, plugin-preflight) - `research-pkg` (Python tooling: uv + ruff + pytest) - `ui-project` (JS/UI/animation/build/test/style) - `static-site` (HTML/CSS/JS, i18n, Lighthouse via chrome-devtools) - `ml-research` (HF Hub, datasets, gpustat, orchestra-ml-skills) - `web-perf` (Lighthouse + a11y tooling) - `image-or-video-work` (sharp, ffmpeg, etc.) - `docs-site` (mkdocs) - `electron-or-desktop` (Electron + electron-builder) 5. **Consumption mode** (single-select): - Raw URL imports — always live, requires network - Local clone (`~/.claude/agent-harness/`) — fast, offline - Plugin install (P10+) — most native; only if `/plugin install jajupmochi/agent-harness` was run 6. **Personal-preference rules** (multi-select): - `output-brevity` — no end-of-batch recap, no echo, prefer Edit over Write - `tool-proactivity` — fire installed tools without asking - `no-reread-files` — trust in-session memory of file contents ## Step 3: Resolve consumption mode Build the `@import` paths based on the user's choice: - **Raw URL**: `@https://raw.githubusercontent.com/jajupmochi/agent-harness/main/<path>` - **Local clone**: `@~/.claude/agent-harness/<path>` - **Plugin**: (no `@import` lines — plugin auto-loads) ## Step 4: Compose the project Based on choices, do: ### 4.1 Copy the template (if a project type was selected) ```bash # Python research: cp -r ~/.claude/agent-harness/templates/research-package-py/. ./ # Static site: cp -r ~/.claude/agent-harness/templates/personal-cite-static/. ./ ``` ### 4.2 Substitute placeholders Ask the user for:
From jajupmochi/agent-harness, MIT.
More from jajupmochi/agent-harness
- 0.502score
claude-desktop-sessions
jajupmochi
Use when Claude Code sessions started in the terminal CLI don't appear in the Claude Desktop app's Code tab, when you want to hand a running CLI session to…
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
doc-writing
jajupmochi
Use when writing or updating any document a human will read — a design or architecture spec, a feature manual and its in-app guide, an audit / bugfix / test…
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
agent-update-watcher
jajupmochi
Use to keep up with the fast-moving agent ecosystem (new/updated CLIs, plugins, skills) WITHOUT burning tokens on constant checks.
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
figma-design-fetch
jajupmochi
Use when the user shares a figma.com URL, wants to implement/mock a UI from Figma, or do design-to-code — connect the Figma MCP, fetch the design…
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
autopilot
jajupmochi
Set up or manage the autopilot daily autonomous project-driver.
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
publish
jajupmochi
Tag a new version of agent-harness, push to GitHub, and create a GitHub release.
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
figma-authoring-constraints
jajupmochi
Use when a designer asks how to structure a Figma file so it produces clean code, when a design keeps yielding pixel-snapshot output, or when…
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
init-codex-config
jajupmochi
Scaffold or migrate a project to use agent-harness from Codex.
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
long-running-tasks
jajupmochi
Manage long-running commands and multi-step work in Codex.
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
agent-config-adapter
jajupmochi
Adapt an existing agent configuration or plugin to a new agent or model route.
SkillCFreshness: C★ 1Claude Code +9 - 0.494score
new-rule
jajupmochi
Scaffold a new workflow rule under rules/<kebab-name>/.
SkillCFreshness: C★ 1Claude Code +9 - 0.494score
new-hook
jajupmochi
Scaffold a new hook under hooks/<kebab-name>/.
SkillCFreshness: C★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Jul 18, 2026 (3 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/jajupmochi/agent-harness
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from jajupmochi
- 0.502score
figma-design-fetch
jajupmochi
Use when the user shares a figma.com URL, wants to implement/mock a UI from Figma, or do design-to-code — connect the Figma MCP, fetch the design…
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
claude-desktop-sessions
jajupmochi
Use when Claude Code sessions started in the terminal CLI don't appear in the Claude Desktop app's Code tab, when you want to hand a running CLI session to…
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
doc-writing
jajupmochi
Use when writing or updating any document a human will read — a design or architecture spec, a feature manual and its in-app guide, an audit / bugfix / test…
SkillCFreshness: C★ 1Claude Code +9 - 0.502score
agent-update-watcher
jajupmochi
Use to keep up with the fast-moving agent ecosystem (new/updated CLIs, plugins, skills) WITHOUT burning tokens on constant checks.
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
figma-authoring-constraints
jajupmochi
Use when a designer asks how to structure a Figma file so it produces clean code, when a design keeps yielding pixel-snapshot output, or when…
SkillCFreshness: C★ 1Claude Code +9 - 0.497score
publish
jajupmochi
Tag a new version of agent-harness, push to GitHub, and create a GitHub release.
SkillCFreshness: C★ 1Claude Code +9
Cite this page
init-agent-config skill. Agent Mart. https://agentmart.dev/skills/jajupmochi-init-agent-config (last updated 2026-10-24)