HotRepo

What HotRepo does

HotRepo MCP server helps agents discover and evaluate open-source repositories that have appeared on GitHub Trending and received a HotRepo review. The service combines repository statistics with one-line or full Markdown reviews, topic classification, trending history, and related Hacker News discussions.

Its coverage is centered on reviewed projects rather than the entire GitHub repository universe. Search returns repositories in HotRepo's reviewed collection, while repository details are available only for projects that have both trended and been reviewed. Each reviewed project belongs to one HotRepo topic based on what the project helps people do.

The service is hosted and exposes read-only operations. It is useful for research, project discovery, ecosystem monitoring, and building agent workflows that need curated context around repositories rather than raw repository metadata alone.

How it works

Trending boards are organized around UTC days. The standard boards rank repositories by stars gained, while HotRepo's new-repository board uses measured Stars/hour. Board entries include HotRepo's one-line review when available. GitHub's displayed order is not treated as an exact stars-gained ranking, so a repository's HotRepo rank and its GitHub position can differ.

Historical ranking fields preserve the best GitHub position observed during the relevant scrapes. Depending on the board, this can represent a repository's best position on that day's board or its best position on GitHub's all-languages daily Trending page up to that date. Unknown historical positions are represented as null.

New-repository data also includes growth timing and the observation window used to calculate Stars/hour. Live comparisons use the preceding successful hourly run, including retained projects; other boards compare dates. Older records may use average stars per day instead.

Tools and capabilities

• get_trending retrieves a daily GitHub Trending or new-repository board, including rankings, growth measurements, and short reviews.
• get_repo returns a repository's complete HotRepo review in Markdown, statistics, topic, daily trending history, and linking Hacker News threads.
• search_repos searches reviewed repositories, with newest reviews first. Use get_repo to retrieve a full result.
• list_topics lists HotRepo topics and the number of reviewed projects assigned to each.
• get_topic lists reviewed repositories in a selected topic, ordered by recent trending activity.
• get_monthly_report returns a monthly editorial report covering major star gainers, notable new repositories, and topic leaders.
• list_favorites lists repositories favorited by the API key's owner, with a one-line review when one exists.

Favorites are managed on project pages at hotrepo.ai, not through an MCP tool. The monthly report and trending tools are suited to time-based discovery, while topic and search tools support narrower research questions.

Setup and configuration

HotRepo is provided as a hosted MCP service and requires a free API key. The key identifies the account whose favorites are returned by list_favorites. Favorites must be added through the HotRepo website's project pages.

Connect the hosted service using an MCP client that supports the service's connection method and provide the API key through that client's configuration.

Limitations and notes

HotRepo's repository-level coverage is limited to projects that have trended and been reviewed. Daily boards from the newer review-filtered period contain only repositories with published reviews, while earlier archived boards can contain entries without one.

GitHub rank fields are observations from scrapes, not a permanent ranking history. Values can be unknown for older dates, and the displayed HotRepo ordering should not be interpreted as GitHub's own ordering. New-board growth rates depend on the measured observation window; legacy records can expose a different average-rate field.

All listed operations are read-only. The server can show account favorites, but it does not add or remove them.