Optifeed Radar is an open-source AI visibility checker used to measure whether a brand gets recommended when buyers ask questions to AI. It is also known as AI visibility, generative engine optimization (GEO), answer engine optimization (AEO), or AI-SEO.
Purpose The tool is aimed at two types of AI agents: on one hand, it measures how AI agents view and recommend your brand; on the other hand, it can also be run by your own AI agent (via CLI, JSON, and MCP server). It checks whether the models behind ChatGPT, Perplexity, Gemini, and Claude actually recommend you, and tells you where you stand compared with competitors.
Key capabilities
- Zero-key audit: no API key required, no AI calls, checks AI crawler access (robots.txt), llms.txt, schema.org structured data, meta basics, and sitemap, and outputs a 0-100 AI readiness score.
- Visibility check: after setting at least one engine API key, it discovers the brand, generates buyer prompt packs, asks the engines questions, and aggregates recommendation, position, and share-of-voice scores into a single AI visibility score. The score only reads buyer questions that do not mention the brand; questions that mention the brand are reported separately as a reputation report.
- Shopping check (beta): performs the same evaluation for a single specified product, with each product receiving an independent 0-100 visibility score, and the report sorted by actual engine behavior.
- Snapshot comparison: get_snapshot_diff can compare changes between two saved runs.
- Supports four engines: OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini), and Perplexity.
- Distinguishes between grounded (citing web sources) and parametric (answering only from model weights) engines, and reports them separately.
How to use
- Run via npx: npx optifeed-radar, no installation required.
- Zero-key audit: npx optifeed-radar audit yourbrand.com.
- Full check: set at least one engine API key (can be placed in a .env file or exported via export), then run the check command.
- Install as an Agent Skill: can be installed directly from the repository, for use by Codex,Claude Code, Cursor, and other compatible AI agents.
- Install as a Claude Code plugin: the plugin bundles the same skill and starts Radar's MCP server from the published npm package.
- MCP server: optifeed-mcp runs via stdio, fetched on demand by npx, with no cloning or building required. It can be configured for Claude Desktop, Claude Code, Cursor, and Windsurf.
- Common check flags: --json, --report report.html, --max-cost, --quick, --grounded, --fail-under, --yes.
Use cases
- Brands want to know whether AI engines recommend them when buyers ask questions.
- E-commerce stores want to know whether AI recommends them in "where to buy" type questions.
- Product teams want to evaluate AI visibility product by product.
- Developers or AI agent workflows need to run visibility checks programmatically.
Limitations
- Users must provide their own API keys for each engine; checks consume the user's own API quota.
- The audit command is free and requires no key; the cost of the check pipeline varies by engine, prompt pack size, and provider pricing.
- Scores are estimates and will fluctuate between runs.
- The shopping check lists at most 10 products each time and only checks the first 10.
- Requires Node 20 or higher.
- The standalone skill and Claude plugin do not create a public ChatGPT app; ChatGPT support will be launched separately after Radar is packaged and tested for OpenAI's plugin and MCP distribution route.
- Data is stored locally, with no Optifeed-hosted backend; keys are sent only to the corresponding AI provider and are not logged.
