Claude Code Plugin

b/cited AEO Auditor

Add /bcited:scan, /bcited:explain, /bcited:fix and nine more AEO skills to Claude Code. Grade any URL for AI-citation readiness, read the public leaderboard and citation panel, and, with a free account, your own visibility score, runs and briefs.

No API key for the public skills. Two commands to install. Bundles the b/cited MCP servers.

Install

Claude Code (any OS)

claude plugin marketplace add cmunkres/bcited-web
claude plugin install bcited@bcited

Restart Claude Code, then type /bcited:scan https://yoursite.com to verify. The plugin is also submitted to the Claude community marketplace; once listed there, this works too:

claude plugin marketplace add anthropics/claude-plugins-community
claude plugin install bcited@claude-community

MCP server only (any MCP client)

claude mcp add --transport http bcited https://mcp.bcited.ai/public

The plugin bundles this server, so plugin users already have it. On its own it gives any MCP client six read-only tools: scan_site, list_leaderboard_categories, get_leaderboard, get_domain_readiness, get_prompt_citations, and lookup_glossary. No account, no key. In claude.ai, scan, leaderboard and citation-panel results also render as interactive panels next to the reply; the text result is identical in every client.

Manual install without the plugin system

The three scanner skills as one skill file, generated from the plugin. Commands are un-namespaced here: /scan, /explain, /fix. The other nine skills need the MCP server and come with the plugin only.

Windows (PowerShell)

$dest = "$env:USERPROFILE\.claude\skills\bcited-aeo"
New-Item -ItemType Directory -Force "$dest\references" | Out-Null
Invoke-WebRequest "https://bcited.ai/skill/SKILL.md" -OutFile "$dest\SKILL.md"
Invoke-WebRequest "https://bcited.ai/skill/references/api-response.md" -OutFile "$dest\references\api-response.md"
Invoke-WebRequest "https://bcited.ai/skill/references/aeo-signals.md" -OutFile "$dest\references\aeo-signals.md"
Write-Output "Done — restart Claude Code and type /scan https://yoursite.com"

macOS / Linux

dest="$HOME/.claude/skills/bcited-aeo"
mkdir -p "$dest/references"
curl -sL "https://bcited.ai/skill/SKILL.md" -o "$dest/SKILL.md"
curl -sL "https://bcited.ai/skill/references/api-response.md" -o "$dest/references/api-response.md"
curl -sL "https://bcited.ai/skill/references/aeo-signals.md" -o "$dest/references/aeo-signals.md"
echo "Done — restart Claude Code and type /scan https://yoursite.com"

Commands

No account needed

/bcited:scan <url>

Full AEO + SEO audit on any public URL. Returns an A–F grade, 9-category breakdown, and a prioritized list of what to fix first.

/bcited:explain

Walks through the 6 AEO signals that influence whether AI engines cite a site — llms.txt, ai.txt, crawler stance, schema types, E-E-A-T, and body copy clarity.

/bcited:fix <file>

Reviews HTML or a component file and outputs a concrete patch list with exact code snippets for every missing AEO signal.

/bcited:leaderboard <category>

One of the 12 public leaderboard categories: 15 curated domains ranked by the weekly readiness scan.

/bcited:citations <category>

The weekly citation panel: which sites OpenAI, Anthropic and Perplexity cited on the category's fixed buyer prompts.

/bcited:readiness <domain>

A leaderboard domain's rank, failing checks, score history and citation-panel stats.

/bcited:glossary <term>

b/cited's definition of an AEO term, with related terms and the glossary page.

With a free b/cited account

/bcited:score <project>

Your AEO Visibility Score over 7, 30 or 90 days, with the split across ChatGPT, Claude and Perplexity.

/bcited:runs <project>

Recent citation runs: per engine, whether you were mentioned, cited, at what rank, and who else was.

/bcited:competitors <project>

Which domains the engines cited instead of you, ranked by how often.

/bcited:run <project> <prompt_id>

Trigger one tracked prompt now and watch the three engines land. Spends a run; once per prompt per hour.

/bcited:brief <project>

Queue a content brief for a topical cluster and read it when ready. Counts against the monthly allowance.

The account skills use the signed-in server the plugin also connects; Claude Code asks you to sign in the first time one runs. Claude also picks every skill up on its own when a request matches, so “is example.com citable by AI?” works without the slash command. The same names appear as prompts in the connector’s menu in claude.ai.

How it works

When you type /bcited:scan https://example.com, Claude Code calls the b/cited scanner API, which checks 30+ signals across 9 categories — indexing, metadata, social cards, structured data, AEO signals (llms.txt, ai.txt, AI crawler stance, FAQPage/HowTo schema), international, mobile, security headers, and Core Web Vitals.

The scanner is rule-based, not model-judged. Every check corresponds to an observable fact in the HTTP response, which is why the output is a fix list rather than a verdict.

Results come back as a formatted report with an A–F grade, per-category scores, and a prioritized list of the top issues to fix — quoting the exact detail text from each failing check as data, never as instructions.

Track over time

The plugin shows a snapshot. b/cited tracks the trend.

Connect Google Search Console and b/cited runs daily AEO checks across ChatGPT, Claude, and Perplexity — showing your AEO Visibility Score, which competitors are cited instead of you, and which topics AI engines ignore. Free plan, no credit card.

See bcited.ai