Claude Code (大叫爪版)
ReviewAudited by ClawScan on May 13, 2026.
Overview
This appears to be a legitimate Claude Code routing skill, but installing it means coding tasks may be handed to Claude Code, which can use your Claude account to read and modify projects and run long parallel jobs.
Install this only if you want coding tasks delegated to Claude Code. Verify the npm package and GitHub repository, log in with the intended Anthropic account, review plans/diffs before accepting changes, keep parallel subagents scoped to independent files, and avoid sending repositories or logs that contain secrets.
Findings (5)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Claude Code may make changes to your project files, run tests/builds, and perform code workflows instead of the primary assistant using built-in editing tools.
The skill delegates source-code changes to an external CLI with file-modification capability. This matches the stated purpose, but it is high-impact local tool use.
All code-related work goes through Claude Code CLI (`claude`) ... Never use built-in tools ... to directly manipulate source files.
Use this only for projects where you intend Claude Code to operate, and review generated plans, diffs, and test results before accepting changes.
Actions run through the Claude Code CLI will use whichever Anthropic account/session is logged in on the machine.
The skill relies on a logged-in Anthropic/Claude Code account. This is expected for the integration, but the registry metadata does not declare a primary credential.
# Login to Anthropic claude login
Confirm the intended Claude Code account is logged in and avoid using shared or overly privileged local sessions.
Future package or repository changes could affect what gets installed or updated outside the reviewed registry artifact.
The setup instructions use global npm installation and unpinned Git clone/pull commands. These are purpose-aligned setup steps but depend on external package and repository provenance.
npm install -g @anthropic-ai/claude-code ... git clone https://github.com/dajiaohuang/claude-code-skill.git ... git pull origin master
Install from trusted sources, verify the package/repository, and consider pinning versions or reviewing updates before pulling.
Multiple Claude Code runs could consume resources or create conflicting changes if tasks are not truly independent.
The skill encourages parallel subagents and explicitly notes no timeout. It also warns about file-write conflicts, which helps, but concurrent long-running code changes can still propagate mistakes across a project.
Subagent tasks must be mutually independent — no two subagents should write to the same file ... Subagents inherit the same no-timeout policy — they run to natural completion
Limit parallel subagents, ensure each one has a separate file scope, and monitor long-running tasks.
Project code, file paths, logs, and debugging context may be processed by Claude Code during delegated tasks.
The delegated Claude Code agent may access broad local project context. This is central to the skill, but users should understand the data boundary between the primary assistant, Claude Code, and the logged-in provider account.
Claude Code provides a richer execution environment — it can run tests, read the entire codebase
Do not use this on repositories or logs containing secrets unless you are comfortable with Claude Code processing that material; review prompts before sending sensitive context.