# githubworth

> githubworth is a GitHub worth calculator that prices real code. Type a username and it counts the code in that profile's public, original repos, estimates their git history from the commit count, and prices both at the rates Pangea publishes for licensing code to AI labs. Every repo gets a figure, a range and a list of buyer checks it meets or misses. Each profile gets a share card and a README badge. A separate page prices private repos from GitHub, GitLab or Bitbucket, in the browser. Owners pay nothing. A buyer pays githubworth a referral fee when someone we send licenses code. githubworth is run by Dataconto.

Last checked: 2026-10-21.

## Rates used
- Pangea: $100 per million tokens of source code and $40 per million tokens of git history. Non-exclusive license, so the owner keeps the code. Rates checked 2026-10-20 on pangea.ai/code-licensing/companies and in pre.dev's published pricing script, which prices repos for Pangea.
- The figure is the githubworth estimate, not an offer. Pangea licenses only private repos that were never public, so a public profile's figure shows what that code would list for if it were private.

## How the estimate works
- Source: counted with pre.dev's published rules. Code files count in full. Docs, config and data files count up to 10 times the code beside them. Dependencies, builds, caches and minified files are left out. Four bytes make one token.
- History: estimated as 45 × source^0.48 × commits^0.88 tokens, fitted on 477 public repos counted in full with pre.dev's script. The range on each repo runs from 0.27 to 3.8 times the estimate, where 8 in 10 of those repos landed. The calibration data is published.
- Buyer checks: 13 per repo (private, not a fork, source code, history, 100+ commits, a year of work, issues, README, tests, CI, container build, more than one contributor, owned outright). Met, failed or unknown. They never change the figure.
- Username pages count up to 30 of the profile's most-starred public, original repos, and say so when the count is partial.
- Private repos are counted in the browser, so githubworth never sees the code.
- Full method: [How the githubworth estimate is made](https://githubworth.com/methodology)

## Pages
- [GitHub worth calculator](https://githubworth.com/): enter a username, get a figure for every repo.
- [Private repos](https://githubworth.com/private): price private code from GitHub, GitLab or Bitbucket in your browser.
- [Source code vs git history](https://githubworth.com/source-vs-history): what counts as each, and when history is worth more.
- [How we get paid](https://githubworth.com/disclosure) · [About](https://githubworth.com/about) · [Privacy](https://githubworth.com/privacy) · [Terms](https://githubworth.com/terms)
- MCP server: https://githubworth.com/mcp (Streamable HTTP, no key, read-only). One tool, `estimate_repo_value`, takes a public repo as owner/name or a github.com link and returns the same figure, range and buyer checks as the site.
- Sister site: [Dataconto](https://dataconto.com) shows which AI buyers would license company records or private code.