lra-code-review-mcp
lra-code-review(基于 LangGraph 的代码审查智能体)的 MCP server。它把 lra-code-review 包装成只读工具,任何 MCP 客户端(Claude Desktop、Cursor、Windsurf……)都能通过 LLM 审查代码。
通过 stdio 暴露 4 个只读工具:
| 工具 | 用途 |
|---|---|
review_project |
全量代码审查,返回 findings 摘要 |
review_diff |
只审 git diff 变更 |
get_finding |
按 id 取单条 finding 的完整证据 |
generate_fix_prompt |
生成单文件的修复任务 prompt(零 LLM token) |
安装
需要 Python 3.11+。
# 临时跑一次
uvx lra-code-review-mcp
# 或安装
pip install lra-code-review-mcp
lra-code-review-mcp
lra-code-review 是依赖项,会从 PyPI 自动安装。
Docker
每次打 tag 都会发布一个容器镜像到 GHCR,无需安装 Python 就能跑:
docker run -i --rm \
-v /absolute/path/to/lra/config.yaml:/config.yaml \
-e LRA_MCP_CONFIG=/config.yaml \
-e LRA_MCP_PROFILE=deepseek \
-e DEEPSEEK_API_KEY="$DEEPSEEK_API_KEY" \
ghcr.io/030603-ccf/dsh-cra-plugin
-i保持 stdio 打开——MCP 传输走 stdin/stdout。- 挂载你的 lra
config.yaml,并把LRA_MCP_CONFIG指向/config.yaml。 - 用
-e(或--env-file)传 API key;绝不要把 key 打进镜像。
配置
server 的一切配置都来自环境变量;工具参数从不携带 API key。
| 变量 | 必填 | 说明 |
|---|---|---|
LRA_MCP_CONFIG |
是 | config.yaml 的路径——复制本仓库的 config.example.yaml 并填自己的 profile |
LRA_MCP_PROFILE |
否 | 该 config 里的 profile 名(默认用它的 default_profile) |
LRA_MCP_RUNS_DIR |
否 | 审查产物的存放目录(默认 ~/.lra-code-review-mcp/runs) |
API key
你的 key 存在环境变量里,由 config 中的 api_key_env 声明变量名:
profiles:
deepseek:
api_key_env: "DEEPSEEK_API_KEY"
三种方式设置这个变量:
- Shell(CLI):
export DEEPSEEK_API_KEY=sk-... - MCP 客户端的
env块(Claude Desktop / Cursor):在 server 的env里加"DEEPSEEK_API_KEY": "sk-..." - Docker:
docker run -i --rm ... -e DEEPSEEK_API_KEY="$DEEPSEEK_API_KEY" ...
要求(继承自 lra):
api_key必须来自api_key_env,不能内联写在文件里。lsp.enabled必须是false。
MCP 客户端配置
让任意 MCP 客户端指向这个 stdio server:
{
"mcpServers": {
"lra": {
"command": "uvx",
"args": ["lra-code-review-mcp"],
"env": {
"LRA_MCP_CONFIG": "/absolute/path/to/lra/config.yaml",
"LRA_MCP_PROFILE": "deepseek",
"LRA_MCP_RUNS_DIR": "/absolute/path/to/runs"
}
}
}
}
- Claude Desktop:
claude_desktop_config.json - Cursor:
.cursor/mcp.json - 绝不要把真实 API key 写进这个文件——它来自
api_key_env指定的环境变量。
工具暴露在 mcp__lra__* 命名空间下(例如 mcp__lra__review_project)。
本地开发
# 1. 安装 lra(源码可编辑安装)
git clone https://github.com/030603-ccf/code-review-agent
cd code-review-agent && pip install -e .
# 2. 安装 lra-code-review-mcp(可编辑 + dev)
git clone https://github.com/030603-ccf/dsh-cra-plugin
cd dsh-cra-plugin && pip install -e ".[dev]"
# 3. 测试
pytest tests -q
说明
- 4 个工具都是只读的:server 从不修改被审查的项目。
- 审查产物(
findings.json、report.md、summary.json)落在LRA_MCP_RUNS_DIR/<run_id>/。
Metadata
Release files for lra-code-review-mcp 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| lra_code_review_mcp-0.1.0.tar.gz | 15.7 kB | With provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| lra_code_review_mcp-0.1.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 30.8 kB
Release files / lra_code_review_mcp-0.1.0.tar.gz
| Download URL | lra_code_review_mcp-0.1.0.tar.gz |
|---|---|
| Size | 15.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
9bb5732aa0f70d04a853bba5033171b5c00c630e1bf6b556e68a823d243420d5
|
|
BLAKE2b-256 checksum How to use checksums |
15dad413f74aebc3fb85003b88ad5b37c89ef053b339143717fa7ac52ea92c13
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 16, 2026.
Transparency logRelease files / lra_code_review_mcp-0.1.0-py3-none-any.whl
| Download URL | lra_code_review_mcp-0.1.0-py3-none-any.whl |
|---|---|
| Size | 15.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
699f6758ee6984c84342cfc7ecf971c3fc30e59b81dc9ea0ee6dbc4081badf6c
|
|
BLAKE2b-256 checksum How to use checksums |
ac5b39051fbfb08e7b1a7a6e2bbd0e42503c8c695badc0f99b5a31651d5365d0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 16, 2026.
Transparency logSecurity for lra-code-review-mcp 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 2 files in this release have been uploaded with provenance metadata.
Attestation for lra_code_review_mcp-0.1.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 16, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.