RankCLI vs squirrelscan
squirrelscan and RankCLI are the two closest tools in this category: both run hundreds of deterministic checks from a free CLI, gate CI on a score or severity, comment on pull requests, expose an MCP server and ship agent skills. squirrelscan renders pages in a cloud browser and sells credits for that; RankCLI applies fixes and opens pull requests, and runs any AI analysis on your own model.
Facts checked 2026-10-14
12 aspects
| Aspect | RankCLI | squirrelscan |
|---|---|---|
| What it is | An SEO and AI-search linter for developers: 280+ checks from the CLI, a GitHub Action and an MCP server, with fixes as pull requests. | "The website QA tool for your coding agent": an open source CLI that audits websites for SEO, performance, security, accessibility, agent experience and other issues, and gives your coding agent exact fixes.[1] |
| Licence and source | Free npm packages under MIT (@rankcli/cli, @rankcli/agent-runtime, @rankcli/mcp-server). The GitHub Action and MCP server repositories are public; the CLI source is not yet published as a repository. | MIT for the CLI and repository (names and logos trademarked). 274 GitHub stars; repository created 2026-01-02; v0.0.112 released 2026-10-13.[1] |
| Price | CLI, Action, MCP server and local fixes are free with no account. Hosted: Free (1 site, weekly), Solo $9/mo, Pro $29/mo ($19 founding for the first 100), Agency $99/mo; every plan runs all 280+ checks. | Local audits unlimited and free when signed out. Cloud: Free (about 5 cloud audits a month, 500 pages each, one site weekly), Pro $19/mo, Team $29 per seat/mo with a 2-seat minimum, Enterprise custom. Cloud audits cost credits: 50 per audit plus 2 per page.[2] |
| Install and run | npx @rankcli/cli audit --url https://your-site.dev. Action integrallis/rankcli-action@v1 (public repository; Marketplace listing pending). MCP: npx @rankcli/mcp-server, 15 tools. Agent skills: npx skills add integrallis/rankcli-mcp-server. Hosted dashboard at rankcli.dev. | Install script, Homebrew cask, or npm install -g squirrelscan; squirrel audit example.com. GitHub Action squirrelscan/audit-action@v1 on the Marketplace. MCP server local (squirrel mcp) and hosted. squirrel skills install adds Agent Skills and Claude Code and Cursor plugins. Hosted reports and scheduling at reports.squirrelscan.com.[3] |
| What it audits | 280+ checks: crawlability, indexability, on-page, content, links, Core Web Vitals heuristics, security headers, structured data, local SEO, social, accessibility, AI readiness. Multi-page crawl, --max-pages N (default 5). | 300+ rules in 22 categories, including 61 accessibility, 30 performance, 25 structured data, 22 security and 24 content rules. Crawl levels quick (25 pages), surface (100) and full (500), with --max-pages.[1] |
| AI search (GEO) checks | AI crawler access tested by fetching as each crawler and comparing with a browser (catches CDN-level blocks robots.txt cannot show), llms.txt, citation-worthiness, answer structure, E-E-A-T, RAG chunk readiness, entity and brand checks. | 17 Agent Experience rules: whether GPTBot and Claude-User get the same content as a browser, AGENTS.md, llms.txt, Markdown responses, API and MCP discovery, licensing and noai signals, pay-per-crawl, response token weight.[1] |
| CI gating | --fail-on error|warning|notice and --min-score N; exit 0 pass, 1 gate failed, 2 bad invocation, 3 site unreachable, so a down site never passes. The Action posts a PR comment and can alert Slack or Discord. README badge on the hosted Free plan. No SARIF output. | --fail-on with score<N, per-category scores, severity>=error, errors>0, warnings>0; exit 2 when a threshold trips. The Action posts a PR summary comment. Badges on every plan from a published report. SARIF not stated.[3] |
| Fixes | rankcli fix applies fixes locally (--dry-run, --only CODES), free. Hosted auto-fix pull requests through the RankCLI GitHub App: 3 a month on Free, unlimited on paid plans, with per-site open-PR limits. | Fix instructions in every finding, and an audit-website skill that has your coding agent map issues to files, fix them in batches and re-audit. Tool-applied fixes or pull requests not stated.[1] |
| Monitoring and history | Hosted: scheduled audits (weekly on Free, daily and on every push on Pro), history, weekly email report, Slack and Discord alerts on Solo and up. Search Console sync is built but not yet enabled. | Scheduled cloud audits on every plan (one site weekly on Free, daily on Pro and Team), email digest per run, webhooks with a Slack relay as the example. Score history on Pro and up. Search Console not stated.[5] |
| AI model use | The checks are deterministic and need no model. Optional AI analysis runs on your model only: Apple Intelligence on-device on a Mac that has it, else Ollama or LM Studio, else your own OpenRouter, OpenAI, Anthropic, Groq, Gemini or DeepSeek key. Nothing goes through RankCLI. | Local audits are deterministic with no login. Cloud audits add AI page analysis and browser rendering; the provider is not stated. --format llm pipes output to your own assistant.[6] |
| Frameworks | Detects 32 frameworks (94% on a 51-site corpus); fix recipes for 11; build-time plugins astro-rankcli, vite-plugin-rankcli and next-rankcli; framework guides at /seo. | No framework-specific rules stated; the entity-map fix guide has sections for Yoast, Rank Math, WordLift, Next.js and Astro; cloud audits detect the technology.[8] |
| Limits worth knowing | No headless browser: pages are audited as served, which is what crawlers see first, so client-rendered content is reported as such rather than rendered. Performance is heuristic, not Lighthouse lab metrics. First public release in 2026. | JavaScript rendering is cloud-only; the local CLI does not render. Signed-out runs default to 25 pages. Cloud features are metered by credits. First commit 2026-01-02, still at v0.0.x.[4] |
Pick squirrelscan when
- •You need pages rendered in a browser before auditing, and a cloud plan to do it.
- •Accessibility depth matters most: 61 rules against WCAG.
- •You want a public score badge from a published report on the free plan.
- •You want per-seat team plans with white-label reports.
Pick RankCLI when
- •You want the fix applied, locally with rankcli fix or as a pull request from the GitHub App, not only described.
- •You want AI analysis that never leaves your machine or your own account: Apple Intelligence on-device, Ollama, or your key.
- •You want framework-specific fix recipes and build-time plugins for Astro, Vite and Next.js.
- •You want hosted monitoring billed per site, not per credit.
See the RankCLI side for yourself, free and without an account:
npx @rankcli/cli audit -u https://your-site.devSources
Every squirrelscan fact above was read from these pages on 2026-10-14. If something has changed since, tell us and we will correct it.