Dataset 02 · Enterprise
Enterprise private GitHub & CRMs
276+ companiesCode · Comms · Docs · DatabasesPre-training
Full private corpora from funded startups that wound down — source code, Slack/Lark history, internal docs, CRM records and databases. Every repository is graded for training value, and every licensing chain is signed directly with the originating founder.
- Repositories
- 276, individually scored
- Contributors
- 1,194 total · up to 36 / repo
- Code volume
- ~28.8M raw · 7.2M logical LOC
- Activity
- ~135K commits · 11.2K merged PRs
- Primary stacks
- TypeScript, JavaScript, Python, Go, Ruby
- Data types
- Code · Slack/Lark · Docs · CRM · DBs
- Rights
- Signed directly with founders
per-repo metadata shipped with the corpus: contributors raw / logical LOC language mix commit volume merged PRs duplication rate CI on PRs deployment tier observability test coverage containerized docs density issue tracking avg. fn. length ── Training Value 0–100