Skip to main content
Context7 automatically refreshes library documentation so that developers always receive accurate, up-to-date context in their coding environment. Here’s how it works.

Automatic Refresh

Every time a library is requested through Context7 — whether via the MCP server or the REST API — the system checks when the documentation was last updated. If it’s older than a threshold based on the library’s popularity, a background refresh is triggered automatically.

Refresh Thresholds

The staleness threshold is based on each library’s popularity rank in Context7:
Popularity RankRefresh Threshold
Top 1003 days
Top 1,00010 days
Top 5,00020 days
All others30 days
Popular libraries are refreshed more frequently because developers rely on them more heavily and they tend to get updated faster.
The refresh is triggered in the background and does not affect the response time of your current request. You will receive the existing documentation immediately while the update runs asynchronously.

Manual Refresh

Logged-in users can manually trigger a refresh for any library directly from the library page on Context7 or via the refresh API endpoint. This is useful when you know a library has just released a new version and want the updated documentation immediately, without waiting for the automatic cycle to complete. Library owners who have claimed their library receive higher refresh rate limits, giving them more control over how frequently their documentation can be updated.

What Triggers a Refresh

A refresh is triggered when all of the following conditions are met:
  • The library has been requested recently
  • The documentation is older than the threshold for that library’s popularity rank
If a library hasn’t been requested recently, it won’t be refreshed — keeping the system efficient and focused on actively used documentation.
Private libraries are not refreshed automatically. To update a private library’s documentation, trigger a manual refresh from the library page.

Supported Library Types

Automatic and manual refreshes work across all supported source types:
  • Git repositories — re-parsed from the configured branch
  • Websites — re-crawled from the base URL
  • OpenAPI specs — re-fetched and re-indexed from the spec source