New-term radar skill
new-term-radar
Three skills and an MCP server find new names of AI models, products and repos before keyword tools list them and size the searches behind them
- 1,676 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Calls X, Hugging Face, GitHub, Reddit and the AIsa and GoAnyAPI services and spends credit on your key per call
- Clones the repository and runs its commands, writing reports, snapshots and a database to a local folder
- Uses AIsa, GoAnyAPI, TrustMRR and Resend keys from environment variables and reads your Search Console token
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/new-term-radarDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Calls X, Hugging Face, GitHub, Reddit and the AIsa and GoAnyAPI services and spends credit on your key per call
- Clones the repository and runs its commands, writing reports, snapshots and a database to a local folder
- Uses AIsa, GoAnyAPI, TrustMRR and Resend keys from environment variables and reads your Search Console token
This is third-party code. Review the repository files before installing.
What it does
New-term radar is a daily report on names that are just starting to surge in AI: new models, products, tools and repositories. It reads trails that appear before keyword-tool data does: words the chosen X accounts used for the first time in 60 days, new entries in Hugging Face's trending list, repositories under 14 days old gaining 200 or more stars a day, and surges of new URLs in the sitemaps of AI tool directories. You can add Search Console, Reddit asks and TrustMRR product revenue. For each name a small model judges whether it names one specific thing or is a common word. The radar then builds a keyword table: which generic searches sit behind the new thing (for example «ai image detector» behind SynthID Detector), how many people search them a day according to Google Ads and Google Trends, and whether page one of the results has room. Each row gets a decision: prioritize, watch or skip. After 14 days the radar checks its finds against Google Trends and counts the hit rate. The repository has three skills: `new-term-radar` (the whole path from collection to table), `ai-listening` (collection for the last day only) and `seo-new-keywords-research` (the table for names you already have). The skills call the `python -m radar` command line, and the MCP server exposes `recent_leads`, `keyword_table` and `run_radar`.
Who it is for. For people who build pages, newsletters, directories or products around new AI terms: SEO specialists of AI tool sites, writers and founders.
Good fit when
- You want a morning list of new model, product and repository names before competitors have pages for them
- You have a few names and need to know which generic searches sit behind them and which is worth a page
- You need to check whether page one has room for a query and see the weakest site in the top
- You want to sort what people ask for on Reddit in the form «is there an AI tool for…» and get jobs with a demand estimate
- You want a report that can run on a schedule in GitHub Actions and arrive by email through Resend
Not a fit when
- You have no AIsa key: without it X, search volumes, Google Trends and the name judge do not work, only the free signals run, and the report comes unsorted
- You need exact search volumes in the report: the radar says a name is new and advises checking volume separately before building a page
- You need a general site SEO audit or keyword research on a long-known topic: OpenSEO in the catalog covers that
- You need Russian search data: the radar uses Google Ads, Google Trends, X and Reddit, with no Wordstat or Yandex among the signals
- You need a public list from TrustMRR or Reddit posts: the service terms forbid republishing, so the data is for your own research only
Example request
Run keyword research on PhotoCraft and SynthID Detector and tell me which searches are worth building pages forLimitations
- The main work needs a paid AIsa key billed per call: by the author's October 2026 figures about $0.003 per X account per run and about $0.18 a day for a table on five names. The example config with 35 accounts costs about $15 a month. The skill must state the cost before a paid run.
- The first `daily` run only builds snapshots, so it shows almost nothing but GitHub breakouts; the report from day two is more useful.
- The author has not yet measured the hit rate per signal because the check takes 14 days; the one seeded example, Strata, hit three days before its Trends rise.
- The new-sites signal failed with an error on 2026-10-12 because of filter validation on the AIsa side; switch it off with `--skip newsites`.
- The table is weaker for model names: the evidence is announcements and benchmarks and the phrases come out generic. Google Ads volumes lag by about two months, and the last two Trends days are not settled.
- The AIsa and GoAnyAPI links in the README and SKILL.md are referral links: signing up through them supports the author. MIT licence.
How to disable. Delete the skill folders or run `/plugin uninstall new-term-radar@new-term-radar` if you installed it as a plugin. The MCP server is removed by deleting the `new-term-radar` block from the config and running `pip uninstall new-term-radar`. The repository clone lives in `~/new-term-radar` together with `config.json` and local snapshots: delete the folder by hand. The GitHub Actions schedule is switched off with the `RADAR_ENABLED` variable.
FAQ
How do I find new AI terms and keywords before competitors?
A new name leaves a five-step trail: a lab post on X or a new entry in Hugging Face's top, a repository taking off on GitHub, directory and competitor pages, first impressions in your Search Console, and only then Google Trends. Keyword tools see the last step; the radar reads the first four every day and uses Trends to check itself.
What does it cost to use, and what is needed besides an AIsa key?
The Hugging Face, GitHub, sitemap and Search Console signals are free. Paid calls go through AIsa. A second key, GoAnyAPI, is optional: it adds Google Trends related queries and a one-country series for about $0.07 a day. It needs Python 3.10 or newer, and the radar has no external libraries.
What do the prioritize, watch and skip decisions mean in the table?
`prioritize` means the query rose in the last 30 days, is still up and gets at least `min_daily` searches a day. `watch` means a smaller rise, too early to read, or steady demand of that size. `skip` goes to fading and small queries. If page one belongs to strong sites, `prioritize` becomes `watch ↓`.
Which tools does the MCP server expose?
Three. `recent_leads` shows collected finds, review verdicts and the hit rate per signal for free. `keyword_table` builds the table for names and costs about $0.05 per name plus $0.09 per call, 40–90 seconds a name. `run_radar` runs `sources` or `daily` and is paid: use cron or Actions for a schedule.
How do I connect the MCP server and where do I point the radar config?
Run `pip install "new-term-radar[mcp]"` and add the `new-term-radar-mcp` command to the client config with the `AISA_API_KEY`, `RADAR_CONFIG` and `RADAR_STATE` variables. Point `RADAR_STATE` at the folder your scheduled runs write to, and `recent_leads` shows their finds.
How is the keyword table different from the list of new names?
A name like SynthID Detector has a short life as a query and belongs to Google, while the job it does belongs to nobody. For every new name the radar finds such generic searches, sizes the demand and the room on page one. For a name it reads the most-liked X posts or the repository README.
Discussion
No questions or reviews yet.
Related
SkillsThe MCP server for RoASr: ad audits, KPIs and competitor intelligence across Meta, Google Ads and Yandex Direct in one place
An unofficial MCP server for Rush Analytics: SERP top, Wordstat, suggestions, rank tracking and AI-brand tracking, 14 tools
Yandex rank tracker skill via Apify
claude-skill-yandex-rank-tracker
A skill that uses a third-party Apify scraper to pull a site's Yandex ranking for a keyword, domain and region
squirrelscan Skills
squirrelscan Skills & Plugins
Skills and MCP for website audits with squirrelscan: SEO, performance, security, accessibility and a code fix loop
Monorepo of Yandex MCP servers: Direct, Metrica, Webmaster, Wordstat, Tracker, 360, Maps, SpeechKit, YandexGPT and AppMetrica
SE Ranking SEO Skills
Claude SEO Skills
A set of agent skills that turn SEO data into finished deliverables: briefs, audits, keyword clusters, schema, sitemap
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| New-term radar (this entry) | no | 1,676 | ||
| SMM Radar MCP | no | key is optional | 0 | social-media signals from Wordstat and Yandex results with no LLM inside |
| last30days-cn | no | 1,941 | collects a topic's discussions from the last 30 days on Chinese platforms and writes a report |
README in short
The description in English and Chinese opens with the working diagram: six signals, the name judge, the keyword table and the 14-day review. Then come a table of signals with the price of each, a sample table from a real run, a breakdown of what the radar cannot read, the `config.json` table, a quick start, a sample report and a section on running in GitHub Actions. The repository has the `radar` package, three skills, a plugin file and `server.json`, network-free tests and a section with measurements and known issues.
README badge
Original SKILL.md
--- name: new-term-radar description: One run from signal to decision. Finds the names of new AI models, products and repositories seen in the last 24 hours (X, Hugging Face, GitHub, sitemaps), then builds a keyword priority table for them, each new name and the generic search terms behind it, sized against Google Ads volume and Google Trends, with prioritize, watch or skip. Use when the user wants today's new AI keywords, an AI trend radar, or asks what new AI terms are worth building a page or tool for. Also takes names the user already has. Runs the new-term-radar command line; needs an AISA_API_KEY. --- # New-term radar The whole job in one run: collect what is new, then work out which keywords are worth acting on. If the user wants only the collection, or only the keyword research on names they already have, the split skills `ai-listening` and `seo-new-keywords-research` do each half. ## Before you run anything 1. Find the radar: `$NEW_TERM_RADAR_DIR` if set, else `~/new-term-radar`; clone https://github.com/zhuyansen/new-term-radar there if missing. Run every command from that directory. Python 3.10 or newer; nothing to install. 2. An AIsa key must be available, as `AISA_API_KEY` or from the AIsa CLI (`aisa login` stores it in `~/.aisa`): the X posts, the search volumes, Google Trends and the models all go through it. Without it, stop and say so.