Claude CodeSAFE
claude-code-mcp
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Claude Code MCP is an implementation of Claude Code as a Model Context Protocol (MCP) server. This project allows you to use Claude Code's powerful software engineering capabilities through the standardized MCP interface.
⚠️ DISCLAIMER ⚠️
Claude Code MCP is a auto-generated project by DevinAI, who was prompted to analyse the Claude Code codebase, and generate an MCP server.
This is a proof of concept that I don't advise anyone to use.
What is Claude Code?
Claude Code is Anthropic's CLI tool for software engineering tasks, powered by Claude. It provides a set of tools and capabilities that help developers with:
- Code generation and editing
- Code review and analysis
- Debugging and troubleshooting
- File system operations
- Shell command execution
- Project exploration and understanding
The original implementation is available as a JavaScript module that defines prompts and tools for interacting with Claude's API.
What is MCP?
The Model Context Protocol (MCP) is a standardized interface for AI models that enables consistent interaction patterns across different models and providers. MCP defines:
- Tools: Functions that models can call to perform actions
- Resources: External data that models can access
- Prompts: Predefined conversation templates
By implementing Claude Code as an MCP server, we make its capabilities available to any MCP-compatible client, allowing for greater interoperability and flexibility.
Features
- Full implementation of Claude Code functionality as an MCP server
- Provides tools for file operations, shell commands, and code analysis
- Exposes resources for accessing file system and environment information
- Includes prompts for general CLI interaction and code review
- Compatible with any MCP client
- TypeScript implementation with full type safety
Installation
# C
c1bb3d00c90dOBSERVED · 2026-10-06Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add claude-code-mcp -- npx -y [email protected]
{
"mcpServers": {
"claude-code-mcp": {
"command": "npx",
"args": [
"-y",
"[email protected]"
]
}
}
}Exposed tools (13)
9 read · 2 write · 2 destructive. Blast radius: 2 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
bash | write | Execute a shell command |
clear | destructive | Clear conversation history and free up context |
codeReview | read | Review code for bugs, security issues, and best practices |
compact | destructive | Clear conversation history but keep a summary in context |
editFile | write | Create or edit a file |
grep | read | Search for text in files |
init | read | Initialize a new CLAUDE.md file with codebase documentation |
listFiles | read | Lists files and directories in a given path |
pr-comments | read | Get comments from a GitHub pull request |
readFile | read | Read a file from the local filesystem |
review | read | Review a pull request |
searchGlob | read | Search for files matching a pattern |
think | read | A tool for thinking through complex problems |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- declared (1 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (2)
clear, compact
@modelcontextprotocol/sdk, @types/node, ts-node, typescript, zod
Gates applied: no_behavioural_pass.
c1bb3d00c90dfull audit observations/trust-audit/mcp-server/auchenberg__claude-code.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-06 | c1bb3d00c90d | SAFE | B | 89 | first audit |
Questions
What is the Claude Code MCP server?
claude-code-mcp
What tools does Claude Code expose?
13 in total: 9 read-only, 2 that write, and 2 that can delete or overwrite (clear, compact). Every one is listed on this page with its risk.
Is Claude Code safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 2 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Claude Code need?
No credential environment variables were found in its source, so it appears to need none.
How does Claude Code run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as claude-code-mcp at 1.0.0.
How current is this page?
The grade is for one exact copy of the source (c1bb3d00c90d), read on 2026-10-06. The repository is watched and re-audited when it changes.