What is new-term-radar?
new-term-radar is an MIT-licensed Python tool for people who build pages, posts or products around new AI terms — SEO for AI tool sites, newsletters and directories. It sits in the SEO & AI Search Visibility category of the AI Productivity & Writing tools and is open-source, written in Python and licensed MIT.
Overview
new-term-radar is an MIT-licensed Python tool for people who build pages, posts or products around new AI terms — SEO for AI tool sites, newsletters and directories. Its premise is that a new name leaves a trail in a predictable order: a lab or a well-followed account says it on X and a model enters Hugging Face's trending list, a GitHub repo takes off, directories add pages for it, your own site gets its first impressions, and only then does Google Trends show the curve. Keyword tools read that last step; the radar reads the earlier ones every day.
Signals are numbered in the report. The free ones are Hugging Face trending models and Spaces, GitHub breakouts (a repo under 14 days old averaging 200 or more stars a day), word surges in the sitemaps you list, and your own Search Console. Paid signals — X posts from a list of accounts, newly registered sites that already rank, Reddit and Stack Exchange requests, sites that just added a payment link, revenue data — go through one AIsa API key, with an optional GoAnyAPI key for Google Trends related queries and one-country daily curves. A model judges which candidates are real names, and each name is re-read three days later and reviewed against DataForSEO Trends after 14 days, so the report tracks its own hit rate.
For the names it keeps, the radar writes a keyword table that pairs each name with the generic searches behind it: estimated searches a day, when the term started rising, last week against the week before, word age, the best market, page-one difficulty and shape, where your own sites rank, and a decision of prioritize, watch or skip. The same engine runs as a command line (`python -m radar daily`), three Agent Skills, an MCP server with `recent_leads`, `keyword_table` and `run_radar` tools, and a scheduled GitHub Actions workflow. Version 0.6.0 was published to PyPI on 27 October 2026.
What it does
- Free signals with no key: Hugging Face trending models and Spaces, GitHub breakout repos, sitemap word surges and Google Search Console
- Paid signals through one AIsa key — X accounts, new sites that already rank, Reddit and Stack Exchange asks, new payment links — each with its per-call cost listed in the README
- A model-based name judge plus a second reading three days later and a 14-day DataForSEO Trends review that records hits per signal
- A keyword table with search volume, trend shape, word age, per-market volume and difficulty, page-one shape, your own rank and a prioritize / watch / skip decision
- A `pains` command that groups complaints from Chrome Web Store reviews, Reddit and tutorial comments into gaps a new tool could fill
- Runs as a stdlib-only Python CLI, three Agent Skills, an MCP server for Claude Code, Cursor, Codex and others, or a GitHub Actions schedule
Getting started
The radar needs Python 3.10 or newer and uses only the standard library. Without an AIsa key the free signals still run, but the report lists every term unsorted because nothing judges which ones are names.
Clone and configure
Copy the example config, then edit it: the X accounts to read, the sitemaps to watch and your topics. Read each site's robots.txt before adding its sitemap.
git clone https://github.com/zhuyansen/new-term-radar
cd new-term-radar
cp config.example.json config.jsonRun the daily report
The first run only builds snapshots, so it reports GitHub breakouts alone; from the second run on every signal reports what changed. `daily` also writes report.json beside the Markdown.
export AISA_API_KEY=...
python -m radar daily --out report.mdSplit fast and slow signals
Run the X and Hugging Face sources every few hours and everything else once a day.
python -m radar sources --out sources.md
python -m radar daily --skip x,hf --out report.mdBuild a keyword table for names you already have
python -m radar keywords "name"
python -m radar leadsUse it from an agent
Install the three Agent Skills, or register the MCP server (here with uv, which needs no prior install). The skills state the cost and wait for confirmation before any paid run.
npx skills add zhuyansen/new-term-radar
# MCP server command
uvx --with mcp new-term-radar mcpCommands and code are distilled from the project's own documentation — always check the official repo for the latest.
When to use it
- An AI tool directory or newsletter deciding which new names deserve a page the day they appear
- Finding the generic search ("ai image detector") behind a branded launch and how much demand it has
- Checking whether a rising term's first page of Google is open to a new site before writing for it
- Asking a coding agent for "today's new AI keywords" through the bundled skills or MCP server
How new-term-radar compares
new-term-radar alongside other open-source seo & ai search visibility tools AI/TLDR tracks, ranked by GitHub stars.
| Tool | Stars | What it does |
|---|---|---|
| OpenSEO | ★ 26.8k | A self-hostable SEO workbench — keyword research, rank tracking, competitor and backlink analysis, site audits and AI-visibility checks — that runs on your own DataForSEO key and exposes the whole dataset to agents over MCP. |
| Open SEO MCP Skills | ★ 8.9k | Eight MIT-licensed SEO and GEO skills for Claude — site audit, keyword research, rank tracking, competitor gap, backlinks, AI visibility — that run on your own Search Console, GA4 and ads data through Ryze's hosted MCP connector. |
| new-term-radar | ★ 2.1k | A daily radar that spots new AI names before keyword tools list them, then sizes the keywords behind them |
| DATA HIT Tools | — | Five browser-based analytics tools: AI Answer Monitor and AI Answer Stability measure how often ChatGPT, Gemini, Perplexity, Claude and Google's AI answers name your brand, alongside GA4 and search-demand charts. |