CiteKit

Open-source AI visibility tooling for teams that want to understand whether answer engines can crawl, cite, and recommend their brand.

CiteKit is the free developer and growth-team hook for the CiteOps ecosystem. It gives you local, reproducible scans for SEO, AEO, GEO, local, and voice-search readiness. CiteOps Cloud is the paid autonomous operator that turns those findings into scheduled monitoring, PR/CMS drafts, verification loops, proof cards, and TraceIntelligence.

What CiteKit Does

• Scans a brand/domain against live provider probes when keys are available.
• Benchmarks crawl-based AI-readiness signals such as llms.txt, robots.txt, sitemap, schema, canonical coverage, pricing/docs/comparison pages, freshness, author signals, and stats.
• Monitors share-of-answer style visibility across prompt/provider panels.
• Diagnoses why a brand is losing AI recommendations.
• Generates safe fix recommendations for answer-first content, schema, llms.txt, proof pages, comparison pages, and crawlability gaps.
• Checks provider/integration readiness without printing raw secrets.

Install

npm install -g citekit-cli

Or run from source:

git clone https://github.com/Keesan12/CiteKit.git
cd CiteKit
npm install
npm run build
node packages/citekit-cli/dist/index.mjs --help

Quick Start

citekit doctor
citekit benchmark --preset founders --max-pages 3 --json
citekit scan --name "Your Brand" --domain yourbrand.com --competitor "Competitor One" "Competitor Two"
citekit monitor --name "Your Brand" --domain yourbrand.com --prompt-count 8

Provider-backed commands use environment variables when available:

OPENAI_API_KEY=
ANTHROPIC_API_KEY=
GOOGLE_GENERATIVE_AI_API_KEY=
PERPLEXITY_API_KEY=
OPENROUTER_API_KEY=

Missing keys are reported as blockers. CiteKit does not fake provider results.

Commands

• citekit doctor: Inspect provider and integration readiness with masked evidence.
• citekit scan: Run prompt generation, provider probes, diagnosis, and fix recommendations.
• citekit monitor: Measure prompt/provider visibility and routing signals.
• citekit probe: Run one prompt against configured providers.
• citekit diagnose: Explain why the brand is winning or losing.
• citekit fix: Generate safe fix drafts from diagnosis output.
• citekit score: Score citation/recommendation outcomes from probe output.
• citekit benchmark: Crawl preset domain sets and score answer-engine foundation readiness.

Repository Structure

packages/citekit-core   TypeScript primitives for crawl, probe, monitor, diagnose, fix, score, proof, and benchmark workflows
packages/citekit-cli    Public CLI wrapper around CiteKit core
examples                Copy-pasteable CLI and Node examples
docs                    User guides, upgrade path, and OSS boundary notes
tools                  Public boundary checks and repo utilities

Examples

npm run verify
node packages/citekit-cli/dist/index.mjs doctor --json
node packages/citekit-cli/dist/index.mjs benchmark --preset market --concurrency 2 --max-pages 3 --json

import { benchmarkFoundation } from "citekit-core";

const result = await benchmarkFoundation({
  name: "Example",
  domain: "example.com",
});

console.log(result.scores.overall, result.topActions);

OSS vs CiteOps Cloud

CiteKit is useful by itself: run local scans, benchmark your site, export JSON, and learn what answer engines can verify.

CiteOps Cloud is for teams that want the autonomous loop:

• Scheduled monitoring across ChatGPT/OpenAI, Claude, Gemini, Perplexity, Google surfaces, and OpenRouter-backed panels.
• Website-platform-aware onboarding for WordPress, WooCommerce, Shopify, Webflow, Wix, Squarespace, GoDaddy, Hostinger, Weebly/Square, custom code, and unknown/static sites.
• Approval-safe GitHub PRs and CMS drafts.
• Public proof cards and scoreboard pages.
• TraceIntelligence audit trails and hashed-only Sansa learning.

Start with CiteKit. Upgrade when you need the agent to keep watching, fixing, verifying, and proving outcomes.

Development

npm install
npm run typecheck
npm run lint
npm run test
npm run build
npm run boundary:public

Public Boundary

This repo must stay OSS-safe. It must not import private CiteOps Hosted app/cloud/runtime code, credentials, customer data, internal specs, or private Sansa/TraceIntelligence implementations.

Run this before every release:

npm run boundary:public

License

MIT