Agent skill

Claude Code Narp Runtime

by Peiiii in Peiiii/nextclaw

当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入,尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。

MITAuto-check passed

Install Claude Code Narp Runtime

skills CLI
$ npx skills add Peiiii/nextclaw --skill claude-code-narp-runtime -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install Peiiii/nextclaw claude-code-narp-runtime --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/Peiiii/nextclaw.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/claude-code-narp-runtime .claude/skills/claude-code-narp-runtime && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
claude-code-narp-runtime
GitHub stars
260
Token cost
~1.6k tokens
SKILL.md length
494 words
Files
2
Skills in repo
70
Repo updated
First seen
Licence
MIT

At a glance

当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入,尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。

  • Works in 5 steps: 确认工作目录和数据目录 → 解析可用 launcher → 生成稳定 shim → …
  • SKILL.md covers 安装边界, 这个 Skill 负责什么, 禁止事项 and Setup 流程, plus 4 more sections
  • Calls claude, npm and node

What it does

Claude Code Narp Runtime is an agent skill from Peiiii/nextclaw. 当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入,尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。

Its SKILL.md is about 1.6k tokens, which your agent loads only when the skill is triggered. The skill folder holds 1 other file (for example `marketplace.json`).

It works with Claude Agent SDK. The repository describes itself as: A human-centered long-term AI partner—not a task-centered assistant. The licence is MIT.

Example prompts

  • “/claude-code-narp-runtime”

Requirements

  • Node.js

Workflow steps

5 steps, taken from the step headings in SKILL.md.

  1. 确认工作目录和数据目录
  2. 解析可用 launcher
  3. 生成稳定 shim
  4. 写入 runtime entry
  5. Probe 和真实冒烟

What it can do on your machine

Read from SKILL.md and the folder at commit d130133. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • claude
    • npm
    • node
    • pnpm

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npm and pnpm, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Claude Code Narp Runtime loads about 1.6k tokens when it runs. Until then it costs about 33 tokens; SKILL.md has 494 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~33
When it runs · the whole SKILL.md, loaded when a task matches
~1.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from Peiiii/nextclaw at commit d130133, republished under its MIT licence (© Peiiii). 494 words, ~1,554 tokens.

Download SKILL.mdSave it as .claude/skills/claude-code-narp-runtime/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
claude-code-narp-runtime
description
当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入,尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。

Claude Code NARP Runtime

使用这个 skill 时,目标不是让用户自己准备 PATH 或手改配置,而是把 Claude Code 接成 NextClaw 统一 runtime registry 里的正式会话类型。

首选产品路径固定为:

  • runtime entry id: claude
  • runtime entry label: Claude Code
  • runtime type: narp-stdio
  • wire dialect: acp
  • model selection: optional(Claude Code 的 Runtime default + NextClaw 模型)
  • runtime launcher: nextclaw-claude-code-narp
  • route ownership: NextClaw -> RuntimeRoute(model/apiBase/apiKey/headers) -> Claude Code NARP wrapper -> Claude Code SDK/CLI

不要把 Claude Code 做成核心系统里的 provider 特判。NextClaw 核心、kernel、service 只感知 narp-stdio,不感知 claude。

安装边界

始终区分三层:

  • 这个 NextClaw marketplace skill:负责用户接入流程、检查、修复和冒烟。
  • Claude Code NARP wrapper:@nextclaw/nextclaw-narp-runtime-claude-code-sdk,提供 nextclaw-claude-code-narp。
  • Claude Code SDK/CLI:由 wrapper 调用,真实执行 agent loop。

安装这个 skill 不等于 Claude Code runtime 已经可用。skill 必须自己检测 launcher,缺失时安装或修复 wrapper,并生成 NextClaw 可稳定调用的绝对路径 shim。

这个 Skill 负责什么

  • 解释 Claude Code 通过 NARP stdio 接入的正确心智模型。
  • 发现、安装或修复 nextclaw-claude-code-narp。
  • 在 ${NEXTCLAW_HOME:-~/.nextclaw}/bin 下生成稳定 shim。
  • 写入或修复 agents.runtimes.entries.claude。
  • 运行 runtime probe。
  • 运行真实 Claude Code 首条消息冒烟;需要能力验收时继续覆盖工具调用和思考事件。
  • 对常见失败给出有界诊断,不把问题推给用户自己猜。

禁止事项

  • 不要告诉用户“先把 nextclaw-claude-code-narp 放到 PATH 里”作为完成条件。
  • 不要要求用户手工编辑 config.json,除非当前 agent 没有文件写权限。
  • 不要通过命令行环境变量开关决定是否走 Anthropic/OpenAI gateway;wrapper 应根据 NextClaw route 的 provider 形状选择通用桥接路径。
  • 不要修改通用 NARP stdio client 来识别 Claude Code。
  • 不要在 core/kernel/service 注入 Claude 默认 entry 或 provider 分支。
  • 没有 probe 和真实模型回复时,不要声称已经完成接入。

Setup 流程

当用户要求“接入 Claude Code”“安装 Claude runtime”“doctor Claude Code runtime”或类似目标时,按下面顺序执行。

1. 确认工作目录和数据目录

先确定 NextClaw 数据目录:

bash
NEXTCLAW_HOME="${NEXTCLAW_HOME:-$HOME/.nextclaw}"

需要写入:

  • $NEXTCLAW_HOME/bin/nextclaw-claude-code-narp
  • $NEXTCLAW_HOME/config.json

如果这是首次接入,先说明 Claude Code runtime 会读取本机工作区并可能执行本地工具;涉及删除、发送、提交、发布等外部可见动作时仍需用户确认。

2. 解析可用 launcher

按优先级解析真实 launcher:

  1. 如果当前在 NextClaw 源码仓库,并且存在已构建文件: packages/extensions/nextclaw-narp-runtime-claude-code-sdk/dist/controllers/claude-code-narp.controller.js
  2. 如果源码存在但 dist 不存在,先构建: pnpm --filter @nextclaw/nextclaw-narp-runtime-claude-code-sdk build
  3. 如果系统已有 nextclaw-claude-code-narp,可作为真实 launcher 来源。
  4. 否则安装 wrapper 包到 NextClaw 管理目录:
bash
mkdir -p "$NEXTCLAW_HOME/runtime/claude-code-narp-runtime"
npm install --prefix "$NEXTCLAW_HOME/runtime/claude-code-narp-runtime" @nextclaw/nextclaw-narp-runtime-claude-code-sdk@latest

安装后解析:

bash
"$NEXTCLAW_HOME/runtime/claude-code-narp-runtime/node_modules/.bin/nextclaw-claude-code-narp" --help

如果包未发布、npm 不可用或安装失败,报告具体 blocker。不要退化成要求用户手动放 PATH。

3. 生成稳定 shim

把 runtime entry 的 command 固定成绝对 shim 路径,而不是裸命令名。

Unix shim 内容形态:

sh
#!/usr/bin/env sh
exec "/absolute/path/to/nextclaw-claude-code-narp-or-js" "$@"

如果真实 launcher 是 .js 文件或没有执行位,则 shim 应改为:

sh
#!/usr/bin/env sh
exec node "/absolute/path/to/claude-code-narp.controller.js" "$@"

写入后执行:

bash
chmod +x "$NEXTCLAW_HOME/bin/nextclaw-claude-code-narp"
4. 写入 runtime entry

用 JSON parser 读写 $NEXTCLAW_HOME/config.json,保留现有配置,只补齐或修复 agents.runtimes.entries.claude:

json
{
  "enabled": true,
  "label": "Claude Code",
  "icon": {
    "kind": "image",
    "src": "app://runtime-icons/claude.ico",
    "alt": "Claude"
  },
  "type": "narp-stdio",
  "config": {
    "wireDialect": "acp",
    "processScope": "per-session",
    "modelSelectionMode": "optional",
    "command": "/absolute/NEXTCLAW_HOME/bin/nextclaw-claude-code-narp",
    "args": [],
    "env": {},
    "startupTimeoutMs": 10000,
    "probeTimeoutMs": 3000,
    "requestTimeoutMs": 120000
  }
}

不要把 provider route、api key、model 写进 runtime entry。模型和 provider 仍由 NextClaw 的 provider/route 配置决定。

Runtime default 是明确的用户选择:选中后不传 NextClaw provider route 或 model,并启用 Claude Code 的 user/project/local settings,让 Claude Code 使用自己的配置、鉴权和默认模型;选择其它模型时继续走 NextClaw provider route,不得用失败重试在两条路径之间自动切换。

Show full SKILL.md (212 more words)Show less
5. Probe 和真实冒烟

接入完成必须同时满足:

  1. agents.runtimes.entries.claude.type 是 narp-stdio。
  2. wireDialect 是 acp。
  3. modelSelectionMode 是 optional。
  4. command 是可执行的绝对 shim 路径。
  5. NextClaw runtime probe 能看到 Claude Code ready。
  6. Runtime default 和至少一个显式 NextClaw 模型都能通过 Claude Code NARP 链路返回真实回复。

开发仓库里优先使用已有 smoke 脚本或 NCP chat smoke;不在源码仓库时,使用当前 NextClaw 实例或 CLI 能提供的最近真实会话路径。验证不是看配置文件长得对,而是看真实会话返回。

能力验收至少分三档:

  • 文本:要求模型回复固定 marker。
  • 工具:要求执行一个安全、可控的本地命令,并确认有 tool-call start/result。
  • 思考:启用对应模型的 thinking/reasoning 参数,确认 NCP stream 有非空 reasoning。

如果用户说“跑通”或“完成”,默认至少跑文本;如果用户明确要求工具/思考或这是 agent runtime 新接入,必须覆盖“思考 + 工具 + 最终文本”的同轮冒烟。

Doctor

当用户要求 doctor 或报错时,按顺序检查:

  1. $NEXTCLAW_HOME/config.json 是否存在且 JSON 可解析。
  2. agents.runtimes.entries.claude 是否存在、enabled 是否为 true。
  3. type、wireDialect、processScope、modelSelectionMode 是否符合合同。
  4. command 是否是绝对路径、文件是否存在且可执行。
  5. shim 指向的真实 launcher 是否存在,--help 是否能启动。
  6. NextClaw runtime list/probe 是否显示 Claude Code ready。
  7. provider route 是否能解析出模型、apiBase、apiKey、headers。
  8. 真实 Claude Code NARP 会话是否能回复。

根据第一个失败点修复。不要在下游事件层伪造成功。

常见问题

  • command_missing:重新生成 shim;如果真实 launcher 缺失,安装 wrapper 包。
  • npm 安装失败或包未发布:说明这是 wrapper 分发 blocker,给出当前源码构建方案或等待发布,不要说接入已完成。
  • provider 鉴权失败:修复 NextClaw provider/route 配置,不要改 runtime entry。
  • 只有文本没有 reasoning:先做 provider 直连、bridge 直测、Claude Code raw event 三段对照,再判断是 provider 参数、bridge 字段形状还是 Claude Code SDK/CLI 暴露问题。
  • 工具调用卡住:检查 Claude Code 非交互 permission 配置和工具权限,不要把 permission 问题改成 NARP client 特判。
  • gateway 选择异常:修复 wrapper 的 route 解析逻辑,不要引入 NEXTCLAW_CLAUDE_USE_ANTHROPIC_GATEWAY 这类用户命令行开关。

移除

删除 runtime entry、shim 或 wrapper 安装目录属于破坏性配置变更。除非用户明确要求“移除 Claude Code runtime”,否则只做 disable/repair,不删除文件。

完成标准

只有在以下条件全部满足后,才能说 Claude Code NARP runtime 已可用:

  • Claude Code 是统一 runtime registry 里的正式 session type。
  • runtime entry 只通过 narp-stdio(acp) 接入。
  • 模型选择同时提供 Runtime default 和 NextClaw 已配置模型。
  • launcher 不依赖用户手动 PATH,而是 NextClaw 管理的绝对 shim。
  • 真实模型回复通过。
  • 如本次目标包含工具或思考,则对应真实冒烟也通过。

© Peiiii, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in skills/claude-code-narp-runtime of Peiiii/nextclaw.

  • SKILL.md
  • marketplace.json

Open the folder on GitHubat commit d130133

Compare with similar skills

Claude Code Narp Runtime next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Claude Code Narp Runtime compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Claude Code Narp Runtime this skillPeiiii/nextclaw260—~1.6kAutomated safety check: PassMIT
Claude APIKocoro-lab/Kocoro4148 repos~4.5kAutomated safety check: PassApache-2.0
Edgeone Makers MigrationTencentEdgeOne/edgeone-makers-tools1.9k1 repos~4.1kAutomated safety check: PassMIT
Helmor Bump Vendorsdohooo/helmor1.3k—~2.1kAutomated safety check: PassApache-2.0
Edgeone Makers AgentsTencentEdgeOne/edgeone-makers-tools1.9k1 repos~5.6kAutomated safety check: NotesMIT
AdhdUditAkhourii/adhd4.4k—~2.8kAutomated safety check: PassMIT

Similar skills

  • Claude API

    Kocoro-lab/Kocoro

    Build apps with the Claude API or Anthropic SDK. An agent skill from Kocoro-lab/Kocoro.

    414 GitHub starsUsed in 8 repos~4.5k tokens
    AI & LLM EngineeringAuto-check passed
  • Edgeone Makers Migration

    TencentEdgeOne/edgeone-makers-tools

    Migrate existing AI agent projects (LangChain, LangGraph, OpenAI Agents SDK, Claude Agent SDK, CrewAI) to EdgeOne Makers platform conventions.

    1.9k GitHub starsUsed in 1 repo~4.1k tokens
    AI & LLM EngineeringAuto-check passed
  • Helmor Bump Vendors

    dohooo/helmor

    Bump or upgrade the pinned versions of Helmor's bundled agent CLIs, SDKs, and supporting binaries — Claude Code + claude-agent-sdk (lockstep), Codex, Cursor SDK, OpenCode, Kimi, Pi, and gh / glab /…

    1.3k GitHub stars~2.1k tokensUpdated 1 mo ago
    Agent WorkflowsAuto-check passed
  • Edgeone Makers Agents

    TencentEdgeOne/edgeone-makers-tools

    This skill guides building AI agent endpoints on EdgeOne Makers — five framework routes (DeepAgents, LangGraph, CrewAI, OpenAI Agents SDK, Claude Agent SDK), platform-injected context.store /…

    1.9k GitHub starsUsed in 1 repo~5.6k tokens
    AI & LLM EngineeringAuto-check: notes
  • Adhd

    UditAkhourii/adhd

    Parallel divergent ideation for coding agents. An agent skill from UditAkhourii/adhd.

    4.4k GitHub stars~2.8k tokensUpdated 2 days ago
    Productivity & AutomationAuto-check passed
  • Web3 Bug Bounty AI Tools

    tradecatlabs/vibe-coding-cn

    A selection guide to AI-driven tools for Web3 bug bounty work, from autonomous web pentesters to smart contract bug finders, with notes on authorization.

    17k GitHub starsUsed in 2 repos~3.9k tokens
    SecurityAuto-check: warnings

More from Peiiii/nextclaw

All 70 skills in this repo
  • Nextclaw Skin Studio

    Peiiii/nextclaw

    A skill your agent uses when a user wants to browse, apply, inspect, create, refine, switch, or remove a NextClaw skin; wants a personal skin with arbitrary CSS, JavaScript, images, SVG, DOM…

    260 GitHub stars~1.8k tokensUpdated yesterday
    Auto-check passed
  • A skill your agent uses when a development task needs visible phase tracing, task-level or phase-level Token measurement, model/effort comparison, or a deterministic usage report or local dashboard…

    260 GitHub stars~802 tokensUpdated yesterday
    Auto-check passed
  • 当 NextClaw 产品更新后需要生成、替换、挑毛病或检查官网、GitHub README、用户文档或社交传播中的真实截图、AI 宣传视觉、整页 HTML 宣传预览、社区二维码等对外视觉资产时使用;也用于“更新截图”“重新截一批图”“做宣传页”“生成 campaign 页面”“视觉审稿”“五星挑刺法”或发布前检查视觉资产。普通站点布局开发或只写文章不触发。

    260 GitHub stars~801 tokensUpdated yesterday
    Auto-check passed
  • UI UX Pro Max

    Peiiii/nextclaw

    A skill your agent uses when the user wants professional UI/UX design guidance, design-system generation, UX review, or stack-specific frontend guidance through a bundled local UI/UX Pro Max dataset…

    260 GitHub stars~2k tokensUpdated yesterday
    Auto-check passed
  • Impeccable

    Peiiii/nextclaw

    A skill your agent uses when the user wants distinctive, production-grade frontend design, anti-generic AI aesthetics, UX critique, technical UI audits, or final polish through bundled Impeccable…

    260 GitHub stars~2.5k tokensUpdated yesterday
    Auto-check passed
  • Curate NextClaw skill resources, including OpenClaw and community sources.

    260 GitHub stars~479 tokensUpdated yesterday
    Auto-check passed

Questions about Claude Code Narp Runtime

What does Claude Code Narp Runtime do?

当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入,尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。. Claude Code Narp Runtime is an agent skill from Peiiii/nextclaw.

How do I install Claude Code Narp Runtime in Claude Code?

Run `npx skills add Peiiii/nextclaw --skill claude-code-narp-runtime -a claude-code`. Or copy the skill folder (skills/claude-code-narp-runtime in Peiiii/nextclaw) into .claude/skills/claude-code-narp-runtime in your project. Claude Code loads it when a task matches its description.

How do I install Claude Code Narp Runtime in Codex?

Run `npx skills add Peiiii/nextclaw --skill claude-code-narp-runtime -a codex`. Or copy the skill folder (skills/claude-code-narp-runtime in Peiiii/nextclaw) into .agents/skills/claude-code-narp-runtime in your project. Codex loads it when a task matches its description.

Can I use Claude Code Narp Runtime in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add Peiiii/nextclaw --skill claude-code-narp-runtime -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/claude-code-narp-runtime, .gemini/skills/claude-code-narp-runtime, .github/skills/claude-code-narp-runtime and .opencode/skills/claude-code-narp-runtime in your project.

What does Claude Code Narp Runtime need to run?

Going by SKILL.md and its folder, Claude Code Narp Runtime needs the command-line tools its instructions call (claude, npm, node and pnpm). Our summary lists: Node.js.

Does Claude Code Narp Runtime access the network?

SKILL.md contains no URLs. Its commands use npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Claude Code Narp Runtime safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Claude Code Narp Runtime use?

Claude Code Narp Runtime is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Claude Code Narp Runtime use?

About 1.6k tokens (SKILL.md is roughly 6.2k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Claude Code Narp Runtime?

Skills that share tags, products or a category with Claude Code Narp Runtime: Claude API (Kocoro-lab/Kocoro, 414 stars), Edgeone Makers Migration (TencentEdgeOne/edgeone-makers-tools, 1.9k stars), Helmor Bump Vendors (dohooo/helmor, 1.3k stars) and Edgeone Makers Agents (TencentEdgeOne/edgeone-makers-tools, 1.9k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Claude Code Narp Runtime?

Peiiii (a GitHub user) maintains it in Peiiii/nextclaw, which has 260 GitHub stars. The repository holds 70 skills in this directory. The repository was last updated on October 6, 2026.

Source: Peiiii/nextclaw on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.