What to do with company code repositories when your software company shuts down

When a software company shuts down, export every repository with its full commit, pull request and review history before billing stops, then decide whether to sell the code with the IP, license the history, open-source it or delete it. Check contractor assignments and third-party code first; established companies can license that history through SourceX.

What does the code record actually include?

A software company's code record is much more than its final source tree. It covers every commit and diff, branches and tags, pull or merge requests, review comments, linked issues, CI results, release notes, design documents and incident postmortems. A plain clone of a repository keeps the commits, branches and tags; pull request discussions, reviews and issues live in the hosting platform and need a separate export.

RecordWhere it usually livesWhat it shows
Commits and diffsGit historyHow the code changed and the author's stated reason
Pull or merge requests and reviewsGitHub, GitLab or BitbucketReviewer feedback, revisions and approval or rejection
Issues and ticketsPlatform issues or JiraThe problem or request behind each change
CI and test resultsBuild pipelines and their logsWhether a change passed, failed or was rolled back
Design docs and decision recordsConfluence, Notion or docs foldersWhy one approach was chosen over another
Incident postmortemsDocs, ticketing and chat exportsWhat broke, how it was found and how it was fixed

Why the history matters more than the final code

AI developers training agents to write, review and fix software need examples of the work itself: a task, an attempt, feedback, a revision and an outcome. A merged pull request with a review thread, a linked ticket and a passing build is one complete example. The final codebase alone shows the answer without the reasoning that produced it.

Depth and discipline matter most. Repositories with years of history, consistent review and changes tied to tickets are worth far more than a large codebase pushed in a handful of big commits. Records from the rest of the company, such as support tickets and product specs, add context; SaaS company wind-downs covers those.

What are the options for the code at shutdown?

OptionWhat it meansWatch out forFits when
ArchiveExport everything to storage the company controls and keep itStorage costs and who holds access after closingAlways, as the first step
Sell with the IPAn asset buyer acquires the code and its copyrightBuyers want clean title and signed contributor assignmentsThe product still has customers or a strategic buyer
License the history through SourceXThe company keeps ownership and grants a defined right to use the history for AI trainingNeeds the qualification baseline and clean rightsAn established company with years of reviewed code
Open-sourcePublish the code under an open licensePublic code cannot support an exclusive training license; third-party code and secrets must be removedGoodwill or community matters more than proceeds
DeleteDestroy it after any retention duties are metThe value is gone for goodRights are unclear or a contract requires deletion

These options can be combined if they come in the right order. Under 17 U.S.C. section 201, copyright ownership can be transferred in whole or in part and any exclusive right can be transferred and owned separately, so a company can license specific rights in its code history while keeping, or later selling, the rest; any sale documents must reflect a license already granted. Open-sourcing last, if at all, keeps the other doors open. Founders weighing a wider sale of startup assets can compare notes in selling shut-down startup data.

Which rights checks come first?

Code is only licensable if the company owns it. Check five things before choosing any option.

  • Employee code: work prepared by employees within the scope of their jobs is a work made for hire owned by the company, as the Copyright Office explains in Circular 30.
  • Contractor code: commissioned work counts as made for hire only in listed categories and with a signed written agreement, so code from contractors and agencies generally needs a written assignment to belong to the company. Find those agreements now, while former staff can still help.
  • Customer-specific code: services contracts sometimes give clients ownership of custom work; leave it out unless the contract says otherwise.
  • Third-party and open-source code: vendored libraries and dependencies stay under their own licenses, and a license from the company covers only what the company owns.
  • Secrets and personal data: API keys, credentials and customer data in fixtures, logs or old commits are removed under redaction rules agreed with the company before any work begins.

This is general information, not legal, tax or financial advice. Confirm with your own counsel, tax adviser or professional body before acting.

How to archive the code before billing stops

  1. Lock deletion rights. Keep at least two owners or admins who will stay through the wind-down, and remove repository deletion rights from departing staff.
  2. List everything. Record every organization, group, repository, fork and archived project, with creation dates and last activity.
  3. Mirror every repository. Clone each one with all branches and tags into storage the company controls.
  4. Export platform data. Pull requests, review comments, issues, wikis and releases sit outside git; use the platform's export or API tools and check its documentation for what each export includes.
  5. Export linked systems. Ticketing, documentation and CI history give the code its context.
  6. Name a custodian. Write down where the exports are, who holds the keys and who can sign for the company after closing; who can sign for a dissolved company helps if the entity is being dissolved.
  7. Only then downgrade or cancel the hosting plan and linked tools.

The data inventory builder helps list the systems and records in one place.

How to tell if a company's code history is deep enough

Use this checklist to judge whether a closing software company's engineering record could interest AI buyers.

  • 50+ full-time employees at peak (contractors excluded), including a real engineering team
  • Several years of commit history in the main product repositories
  • Pull requests with substantive review comments, not routine self-merges
  • Changes linked to tickets that describe the problem
  • CI or test results attached to changes
  • Design documents, decision records or postmortems
  • Code written by employees, or by contractors with signed assignments
  • Someone who can still export, and an authorized sponsor such as the owner, CEO, CFO or another authorized representative

A small team with a short history is unlikely to meet this baseline; proper archiving and an IP sale are the realistic paths there. If the company is already in bankruptcy or an ABC, a trustee or assignee makes the call instead, as can a bankrupt company license its data explains. The full criteria are on who qualifies.

Next step

Founders and advisors can screen the company with the company fit checker, and the company can apply directly at sourcex.si/apply. If you advise closing software companies, register as a partner to introduce them.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
  4. Step 4You get your rewardYour share of SourceX fees becomes payable.

Common questions

Is a git clone enough to preserve a company's codebase?

No. A mirror clone keeps commits, branches and tags, but pull request discussions, review comments, issues, wikis and releases are stored by the hosting platform rather than in git. Those need the platform's own export or API tools. Without them, most of the reasoning that makes a code history useful for AI training is lost when the account closes.

Can we open-source the code and still license the history?

Usually not for an exclusive AI-training license, because public code can be collected by anyone and the exclusivity a buyer pays for disappears. If open-sourcing matters to the founders, sequence it after any license is agreed and only where the license terms allow it. Check third-party code obligations and remove secrets before publishing anything.

Do former employees or contractors need to approve a license?

Generally not for code employees wrote within their jobs, which the company owns as work made for hire. Contractor and agency code is different: the company owns it only through a written assignment or a qualifying signed agreement. Code covered by neither should be excluded, or its rights cleared, before it is included in any license.

What about secrets and customer data buried in old commits?

They must be found and removed before delivery. API keys, passwords, tokens and any customer data in test fixtures, logs or configuration files are handled under redaction rules agreed with the company before work begins. Rotating or revoking old credentials at shutdown is good practice in any case, whatever happens to the code afterwards.

Our startup is small. Is licensing the code history still an option?

Only if the company meets the baseline: 50+ full-time employees at peak (contractors excluded), several years of documented operations, rights to the material and an authorized sponsor. Smaller and younger companies are better served by archiving the code properly and exploring an IP or asset sale. A preliminary fit check needs no contact details and costs nothing.

Free resources

By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09

Know a US company with valuable proprietary data?

Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.

Refer a company →

I own a business

Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.

Start an assessment