Private code without issue history: the referral questions to ask

Short answer

A private code archive shows what changed. Four other records explain it: the issue or requirement says why the change was needed, the pull request and its review show what was accepted and what was pushed back, the tests show the change worked, and the links between them tie each answer to the right change. Before introducing a company with code history, find out which of these it kept and whether they point at each other.

Visual briefFull-size diagram
Four linked cards, issue, patch, review and test, each labeled with the question it answers, with a bracket showing that a code-only archive keeps just the patch.
Illustrative change at a fictional scheduling software company. The links between cards matter as much as the cards.

Which issue, review and test records give context to a private code archive?

Each record type answers a different question, and none can stand in for another. A diff cannot tell you what the customer asked for. An issue cannot tell you whether the fix was correct. A review cannot tell you whether the tests passed.

EvidenceTypical sourceQuestion it answersLost without it
PatchCommits, merged diffsWhat exactly changed in the code?Nothing to evaluate
RequirementIssue, ticket, bug report, specWhy was the change needed, and what counted as done?The task: a change with no stated problem
ReviewPull request comments, approvals, requested changesWas the change acceptable, and what alternatives were rejected?Judgment: which solutions a senior engineer would refuse
TestTest files added or changed, CI resultsDoes the change work, and did it break anything?Proof: no way to check a proposed fix
LinkIssue keys in commits, linked pull requestsWhich requirement, review and test belong to which patch?All of the above become separate piles

The public benchmark SWE-bench shows how much weight the links carry. Its task instances come only from pull requests that are linked to an issue and that change at least one test file. Tests that fail before the change and pass after it are its main signal for whether a fix is correct. A change without a linked issue or a test was simply not usable as a task.

One change, record by record (Illustrative)

A fictional 200-person company sells scheduling software to field service businesses. A customer reports that technicians whose shifts cross midnight are sometimes double-booked.

  1. Requirement. A support ticket escalates to an engineering issue: "Shifts spanning midnight allow overlapping jobs." It includes steps to reproduce and the customer's account tier, which a buyer would never see.
  2. Patch. An engineer changes the overlap check to compare full timestamps instead of times of day, about 40 lines across two files.
  3. Review. A senior engineer requests changes: the first version ignored daylight saving transitions. The second version passes review with one approval.
  4. Test. The pull request adds three tests, one of which fails on the old code and passes on the new.
  5. Link. The pull request description says "Fixes" followed by the issue number, so the issue closes when the change merges.

Take away any one step and the record loses something specific. Without step 1, the change has no stated problem. Without step 3, nobody knows the daylight saving trap existed. Without step 4, there is no proof. Without step 5, someone would have to guess which issue the patch fixed.

Step 5 depends on habit. GitHub's documentation describes two ways a pull request becomes linked to an issue: a closing keyword in the description, which only takes effect when the pull request targets the default branch, or a manual link added from the sidebar. Teams that never do either still have issues and code, just not joined.

What to ask the company, at metadata level

You are establishing what exists, not looking at it. The CTO or head of engineering can answer these from memory or a quick check.

  • Where do issues live, and since when? Code host issues, a separate tracker, or support tickets only.
  • Roughly what share of merged pull requests reference an issue? An estimate is enough. Teams that require a ticket key on every merge answer this quickly.
  • Is review required before merge? If yes, review comments exist for most changes.
  • Do changes usually add or update tests, and are CI results kept? Ask how long CI history is retained, since build logs often roll off quickly.
  • Was history kept through migrations? A move between code hosts or trackers sometimes keeps the code but drops issues and review threads. Our note on whether git history matters covers why.
  • Who owns the code? A product company usually owns its repositories. An agency's client work usually belongs to the client; see who owns the code a development agency writes.
  • Who can authorize a licensing discussion? Usually the owner or CEO, with the CTO and counsel involved.

The metadata-only GitHub inventory template is a convenient way to record answers without touching any repository.

The company must also have 50 or more full-time employees at peak (contractors excluded), a documented operating history, rights to license, and an authorized contact. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English. See who qualifies.

Limits and unanswered questions

  • Secrets and personal data in history. Old commits can contain credentials and customer details. Handling them is part of SourceX's rights review and data preparation, but the company may want to check first; see scanning git history for secrets.
  • Third-party and open-source code. Vendored libraries and copied snippets carry their own licenses. What the company can license is settled in review.
  • Quality is not visible from the outside. A large archive with thin issues may be worth less than a smaller one with rich reviews and tests. Nobody can tell which from a conversation.
  • Separate payments. Any company compensation is set in its own agreement with SourceX. Your reward is paid separately by SourceX and does not reduce it.

For the wider category, see software engineering workflow data and test case data. The owner may also find our questions to ask before licensing company data useful. More guides are in the guides hub.

Next step

Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

If a software company tells you its merges reference issues, require review and add tests, register, confirm the basics on who qualifies, and read the terms before you introduce the owner.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Is a code archive with no issue tracker worth introducing?

It can still be, but it is a weaker conversation. Commit messages and tests carry some context. Say plainly what is missing and let SourceX's review decide; do not oversell it.

What if issues live in a different tool from the code?

That is common, for example Jira for issues and a separate code host. What matters is whether commits or pull requests reference issue keys consistently, so the two can be joined.

Does the company have to hand over its product's source code?

Nothing is decided at introduction. What could be licensed, what stays excluded, and how sensitive material is handled are settled in the company's own agreement and rights review with SourceX.

Sources

  1. swebench.com/original.html
  2. docs.github.com/en/issues/tracking-your-work-with-issues/usi…
  3. refer.sourcex.si/terms

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment