Run a real code review in 10 minutes.

Test the free, local MCP Code Review Server against a deliberately unsafe sample. You will see the finding, the non-zero CI exit code, and how a shared rule profile works. No source code leaves your machine and running the trial creates no payment.

Start the trialSee team options

What you need

A terminal and Python tooling. The commands below use the current PyPI release, 0.1.2, and the sample files already stored in the repository.

Download the standalone trial bundle (sample, shared JSON profile, and both README files) if you do not want to clone the repository.

Install the server

The recommended isolated runner is uvx. The package name and executable name differ, so --from makes the source package explicit.

uvx --from aicraft-code-review --with "mcp<2" mcp-code-review

Or install into an existing environment:

pip install "aicraft-code-review==0.1.2" "mcp<2"

Download the trial files

Clone the repository, then enter the trial directory. The adjacent JSON profile is discovered automatically.

git clone https://github.com/GoodJobwilliam/aicraft.git
cd aicraft/products/mcp-code-review/trial

No Git? Download the files directly:

mkdir aicraft-trial && cd aicraft-trial
curl -O https://aicraft.vip/products/mcp-code-review/trial/sample.py
curl -fsSLo .mcp-code-review.json https://aicraft.vip/products/mcp-code-review/trial/trial-config.json

sample.py · trial-config.json

Run the sample review

The sample intentionally contains a command-injection pattern. A non-zero exit code is expected and is suitable for a merge gate.

mcp-code-review review-file sample.py

Try your own rule

Copy .mcp-code-review.json into a test repository and edit custom_rules. Commit the JSON path and point every teammate to it with MCP_CODE_REVIEW_CONFIG.

mcp-code-review review-file path/to/file.py

Expected result

  • High: command injection from os.system
  • Medium: team-convention rule from the trial profile
  • Exit code 1 when blocking findings are present
  • Report stays local and is readable in CI logs

Decide after one run

  • Did it catch a real issue before merge?
  • Which rule should every repository share?
  • Which result was noisy or missing?
  • Would monthly rules, CI refreshes, or rollout help save time?

Run the same check on a pull request

Share structured trial feedback on GitHub

Prefer email

Ready to test with a team?

Use the public GitHub form and include team size, languages, current review workflow, and the pain you want to validate. Do not include source code or secrets.

Apply for a free team trialPrefer emailGet Team Rules Pack: $49