Skip to main content
Glama
Sign In
coderocket-mcp

coderocket-mcp

by mlgraham
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Remote

@mlgraham/coderocket-mcp

MCP server for CodeRocket Deploy — AI-generated CI/CD workflows and AI code review, available as tools inside Claude Code.

Setup

Claude Code will never prompt you for your API key. This server reads it from the environment once at startup. Set it before you start Claude Code, or the tools will report that they are not configured.

1. Create an API key

Go to deploy.coderocket.com/settings → API Keys → Create Key. The key is shown once — copy it before leaving the page. If you lose it, you must revoke it and create a new one.

2. Export it to your shell profile

echo 'export CODEROCKET_API_KEY=crk_your_key_here' >> ~/.zshrc   # or ~/.bashrc
source ~/.zshrc

3. Restart Claude Code

Required. A running session will not pick up a newly exported variable.

4. Install the plugin

claude plugin marketplace add mlgraham/coderocket-plugin
claude plugin install coderocket@coderocket-marketplace

Then verify with /coderocket:status.

Related MCP server: mcp-agent-review

Tools

Tool

Purpose

deploy_repo

Generate a CI/CD workflow for a repository (polls until complete)

create_pr

Open a pull request with a generated workflow

list_repos

List your connected repositories

repo_details

Details and analysis status for one repository

list_reviews

List AI code reviews

get_review

Fetch a single review with its inline comments

account_status

Account tier, usage, and limits

generation_feedback

Report whether a generated workflow worked

health_check

Check API connectivity and that your key is valid

Configuration

Variable

Required

Default

Purpose

CODEROCKET_API_KEY

yes

—

Your crk_ API key

CODEROCKET_API_URL

no

https://deploy.coderocket.com/api/v1

Override the API endpoint

CODEROCKET_ALLOW_CUSTOM_API_URL

no

—

Set to 1 to permit a non-default host

CODEROCKET_API_URL is validated before use. Because every request carries your API key as a bearer token, the server accepts only the official host or loopback by default, requires HTTPS off loopback, and refuses to start on an unsafe value rather than starting up and sending your credentials somewhere unintended. Pointing at a self-hosted backend requires CODEROCKET_ALLOW_CUSTOM_API_URL=1, which also prints a warning naming the host.

Troubleshooting

Every tool says the key is not set. The variable was not present when Claude Code started. Confirm with echo $CODEROCKET_API_KEY in a new terminal, then fully restart Claude Code — not just the conversation.

"API key has been revoked." The key exists but was revoked. Create a new one in Settings, update your shell profile, and restart Claude Code.

"is not an allowed CodeRocket host." CODEROCKET_API_URL points somewhere unexpected. If you did not set it deliberately, unset it — something in your environment did, and it would have received your API key. If it is an intentional self-hosted backend, set CODEROCKET_ALLOW_CUSTOM_API_URL=1.

Links

License

MIT

Available Tools

9 tools
account_statusA

Get account information including subscription tier, usage, and limits.

Inputs: no named parameters.

create_prA

Create a GitHub pull request with the generated CI/CD workflow.

Inputs: generation_id (required).

deploy_repoA

Analyze a repository and generate a CI/CD workflow. Handles the full async flow internally (analyze → poll → generate → poll) and returns the complete YAML + explanation.

Inputs: repo_id (required), deploy_target.

generation_feedbackA

Submit feedback on a generated workflow (worked, partial, or failed).

Inputs: feedback (required), generation_id (required), comment.

get_reviewA

Get detailed code review with all inline comments organized by severity.

Inputs: review_id (required).

health_checkA

Check connectivity to the CodeRocket API. Useful for debugging setup issues.

Inputs: no named parameters.

list_reposA

List connected repositories with their analysis status, language, and framework.

Inputs: no named parameters.

list_reviewsB

List AI code reviews, optionally filtered by repository.

Inputs: repository_id.

repo_detailsA

Get comprehensive repository details: analysis results, readiness score, optimizations, and generation history.

Inputs: repo_id (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    A3.8/5.0

    Scored across 9 tools

    Disambiguation5/5

    Each tool targets a distinct resource or action: repo listing vs. repo details, deployment generation, PR creation, review listing vs. details, account status, health check, and feedback submission. There is no overlap, and the boundaries between list_repos and repo_details are clear (summary vs. comprehensive).

    Naming Consistency3/5

    The tool names mix verb-first patterns (list_repos, create_pr, deploy_repo, list_reviews, get_review) with noun-first patterns (account_status, health_check, generation_feedback, repo_details). This inconsistency, while not causing confusion, prevents a predictable naming convention.

    Tool Count5/5

    With 9 tools, the server is well-scoped for its purpose—CI/CD workflow generation, PR creation, code reviews, and account management. Each tool serves a clear function, and the count is within the ideal range.

    Completeness4/5

    The core lifecycle is covered: analyze repo, generate workflow, create PR, submit feedback, and access reviews. Minor gaps exist, such as no direct listing/updating of individual workflows and no way to comment on or resolve review comments, but these are non-critical and the surface is largely complete.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Resources

    Related MCP Connectors

    Related MCP Servers