Ship code, get ranked. The SEO and AI-search audit that runs where your code runs: in the terminal, in CI, and inside your AI assistant. 280+ technical SEO and AI-readiness checks, a non-zero exit code when something regresses, and fixes that land as code.
npx @rankcli/cli audit --url https://your-site.devNo account. Nothing leaves your machine except the requests to your own site.
| Package | What it is | Install |
|---|---|---|
packages/cli |
The rankcli command: audit, gate, fix, apply |
npm i -g @rankcli/cli, brew install integrallis/tap/rankcli, scoop install rankcli |
packages/agent-runtime |
The audit engine every surface runs: the checks, the scoring, framework detection, fix recipes | npm i @rankcli/agent-runtime |
packages/mcp-server |
A local MCP server with 15 tools for Claude, Cursor, Codex and any MCP client | npx @rankcli/mcp-server |
packages/github-action |
The GitHub Action that fails the build on SEO regressions and comments on the PR | uses: integrallis/rankcli-action@v1 |
packages/astro-plugin, next-plugin, vite-plugin |
Build-time SEO lint for Astro, Next.js and Vite | astro-rankcli, next-rankcli, vite-plugin-rankcli |
The source is published under the Functional Source License (FSL-1.1-MIT; the plugins are MIT), see License. The hosted product at rankcli.dev (scheduled audits, history, auto-fix pull requests, team reports) is optional and lives elsewhere; the CLI, the engine, the MCP server and the Action are complete without it.
# Audit one page and print every finding, grouped by severity
rankcli audit --url https://your-site.dev --max-pages 1
# Gate a deploy: exit 1 on any error, or when the score is under 80
rankcli audit --url https://preview.your-site.dev --fail-on error --min-score 80
# Machine-readable, with the fix for each finding
rankcli audit --url https://your-site.dev -o json > audit.json
# Preview the code-level fixes for the current project
rankcli fix --url https://your-site.dev --dry-runWhat an audit covers: crawlability and indexing, on-page and structured data, Core Web Vitals signals, mobile, images, security headers, internal linking, content quality, framework-specific rendering problems (32 frameworks detected, 11 with fix recipes), and AI readiness: whether GPTBot, ClaudeBot and the search crawlers can reach the page, llms.txt, answer-first structure, freshness and citation signals. The AI Readiness score measures hygiene on evidence-backed levers; it does not predict citations, and no published score does.
Node 20.19+ (or 22.12+) and pnpm 8.
pnpm install
pnpm build # every package, in dependency order
pnpm test
node packages/cli/dist/index.js audit --url https://example.comThe CLI bundle externalizes the engine, so after changing packages/agent-runtime rebuild it
before the CLI sees the change. CONTRIBUTING.md has the layout, the test conventions and how
a check is added.
A GitHub release tagged vX.Y.Z publishes @rankcli/agent-runtime, @rankcli/mcp-server and
@rankcli/cli to npm, builds standalone binaries for macOS, Linux and Windows, updates the
Homebrew tap and the Scoop bucket, and attaches a Claude Desktop extension (.mcpb).
The changelog is at rankcli.dev/changelog.
The engine, the CLI, the MCP server and the GitHub Action are under the Functional Source License, Version 1.1, MIT Future License (FSL-1.1-MIT): read it, run it, change it, redistribute it, embed it in your own tools and pipelines, all free; the one thing it does not permit is using it to build a product or service that competes with RankCLI. Each version becomes MIT two years after its release. The three framework plugins are MIT. Copyright (c) 2026 Integrallis Software, LLC.