Privacy policy
Last updated 2026-10-22.
This policy is written for the actual SellCodebase workflow: GitHub identity, selected-repository GitHub App access, estimate previews, codebase submission review, encrypted archive storage, analytics, email, optional Slack alerts, payout setup, and account deletion.
This notice describes Claw Solutions, Inc., operating SellCodebase, and the commercial licensing workflow for expressly accepted snapshots under commercial terms v1.1.0. The 24 June 2026 privacy policy remains archived. Publishing this notice does not authorize new uses of historical submissions or replace an earlier agreement.
A new license requires affirmative, versioned acceptance for the identified source. Sign-in, a GitHub connection, an estimate, silence or a policy-page edit grants no commercial rights. All project/path exclusions and cancellation, rejection and deletion controls remain effective. An attestation, NDA or automated scrub does not create missing ownership or a lawful basis to disclose personal data; unresolved authority or privacy issues keep the affected material on hold.
1. Scope
Claw Solutions, Inc., operating SellCodebase (“we,” “us,” or “our”), operates sellcodebase.com and the related repository appraisal, submission, review, payout, and support services (the “Service”). This Privacy Policy explains what we collect, why we use it, who can receive it, and what choices you have.
Your use of the Service is also governed by our Terms of Service. Do not submit source code, files, secrets, personal data, regulated data, or third-party material that you are not authorized to disclose and license.
2. Account and GitHub data
When you sign in or connect GitHub, we collect and store:
- GitHub identity data from OAuth, including GitHub account ID, login/name, email address, avatar URL, provider account ID, provider name, OAuth scopes, token expiration fields, and encrypted OAuth access, refresh, or ID token values when GitHub provides them.
- SellCodebase account data, including user ID, name, email, profile image, role or admin flags, ban fields if applicable, timestamps, authentication sessions, session tokens, session expiration, IP address, user agent, verification records, and database-backed rate-limit records.
- First-party authentication cookies and a short-lived session cookie cache used by Better Auth to keep you signed in and protect account routes.
- GitHub App installation records, including installation ID, GitHub account login, GitHub account ID, account type, repository selection mode, granted permissions, and timestamps.
GitHub OAuth is used for identity. Private repository access uses the SellCodebase GitHub App, not a broad OAuth repo scope. GitHub App installation tokens are generated server-side when needed and are not stored as permanent database fields.
3. Repository browsing and estimate previews
If you type or paste a repository, the Service may parse the repo input and check GitHub for public/private visibility. Signed-in repository screens list the private repositories you selected in the GitHub App, including repo name, visibility, language, size, pushed date, installation summary, and whether a repo already has an active submission.
Estimate previews are intentionally not the persistence boundary for repo appraisal data. A private estimate may read GitHub language bytes through your selected-repo GitHub App access and return an estimated line count, payout preview, quality score, and language breakdown. That route does not create a submission, store the repo estimate, or save the repo URL, score, estimate, or language breakdown in SellCodebase database rows. We store those appraisal details when you submit the repository for review.
4. Submission and review data
When you submit a repository for review or authorize a new snapshot for commercial licensing, we collect and store:
- Repository metadata such as full name, GitHub URL, default branch, submitted commit SHA, GitHub App installation ID, repository size, visibility, active-submission status, and submission timestamps.
- Appraisal and processing results such as estimated and final line counts, internal line-count output, quality scores, quality grades, score reasoning, commit-timing evidence, review-signal packets, estimated payout, final payout, processing errors, and status history.
- Intake answers and flags covering your relationship to the repo, rights and license status, copied or generated code, AI tool usage, model families, project status, usage, revenue range, tests, CI, docs, deployment targets, dataset context, private dependencies, secret handling, personal or regulated data risk, excluded paths, desired offer, deal shape, payout preference, and free-form notes.
- Contact and profile details you provide, such as contact email, phone number, preferred contact method, timezone, preferred deal shape, payout method, PayPal email, Wise email, or bank-transfer account reference. The current in-app bank flow stores an account reference only; routing details are collected later if needed after qualification.
- Admin review fields such as status, rejection reason, admin notes, reviewed timestamp, paid timestamp, submission events, archive state, and archive deletion or failure state.
- For new commercial grants, an immutable acceptance receipt identifies your account, legal owner/payee, named signatory or authorized representative, authority and required-approval declarations, transactional email, acceptance time, commercial terms version and exact text SHA-256 hash, repository and submitted commit SHA, approved related-data categories, exclusions and clearance/hold state. Repository access alone is not ownership evidence.
- Commercial cash-receipt and payment statements record unique receipt/component references, actual cash arrival, attributable Platform receipts, raw/task classification and substantive task evidence, source allocation, rate and contributor amount, partner arrangement and processing-cost disclosures, hold end and actual payment reference/time. Repeated receipts preserve prior earned balances; missing payout setup does not erase debt.
5. Source code access and archives
When a repository is submitted, our ingestion job creates a short-lived GitHub App installation token, downloads the submitted commit tarball to temporary server storage, extracts it, runs the internal TypeScript line counter, scores the codebase, collects review signals, encrypts the tarball locally, uploads only the encrypted archive to Cloudflare R2, and deletes the temporary working files.
Submitted archives use AES-256-GCM with a random per-archive data key. The data key is encrypted before it is stored in Postgres. The database stores archive metadata such as storage provider, bucket, object key, plaintext and ciphertext SHA-256 hashes, byte sizes, encryption algorithm, encrypted data key, IV, auth tag, archive state, stored timestamp, requested-deletion timestamp, and deleted timestamp.
Admin code review can also read selected text files live from GitHub using a short-lived GitHub App token. The admin browser is limited to a bounded file list and blocks large or binary files from inline display. It does not expose GitHub tokens, R2 keys, or archive encryption material to reviewers in the browser.
6. Analytics, logs, email, and Slack
In production, we use Vercel Web Analytics for page-view and product-event measurement. Before page views are sent, the app redacts repo appraisal paths and the repo, redirect, and error query parameters.
Product events are separate from page-view URLs. Submission, cancellation, status-change, GitHub sign-in, bulk-action, and payout events can include operational properties such as repo full name, owner, repo name, submitted URL, default branch, visibility, repo size, estimated lines, payout estimates, score, grade, submission ID, old and new status, coarse failure reason, final payout, whether a rejection reason exists, method selected for payout, field count, and a short hash of your user ID. Payout events do not send the actual payout account details.
We may keep operational logs for debugging, security, availability, and abuse prevention. Logs can include request metadata, user IDs, installation IDs, repo names, status values, error messages, and failure reasons. If email is configured, Resend receives the email address, subject, and message content needed to send seller submission, qualification, rejection, and payment-status emails, as well as internal new-submission review notifications. If Slack alerts are configured, Slack can receive a new-submission alert with repo name, seller email, estimated lines, score, and estimated payout.
7. How we use data
We use collected data to:
- authenticate users, maintain sessions, secure accounts, enforce same-origin and fresh-session checks, prevent abuse, and operate rate limits;
- connect selected GitHub App installations, list selected repositories, verify repository access, and refresh stale installation records;
- provide estimate previews, create submissions, process code snapshots, count lines, score codebases, prepare review evidence, and manage submission status;
- perform manual review, confirm seller authority, evaluate transaction fit, communicate with sellers, prepare diligence, and coordinate potential buyer outreach under the Terms of Service;
- send operational emails, internal alerts, support responses, security notices, and payout-status updates;
- store payout preferences, support payment operations, handle tax or compliance follow-up, investigate fraud or rights issues, and enforce legal agreements;
- for expressly accepted, cleared source scopes only, securely evaluate, scrub, annotate and combine approved material, create substantive task products, and license or commercially deliver it for AI training, fine-tuning, reinforcement learning, testing, evaluation and research to customers or authorized independent commercialization partners;
- measure product usage, diagnose failures, improve flows, and verify that privacy-sensitive paths are being redacted where intended.
8. Who receives data
We do not sell contact or payout details for advertising, and we do not run advertising trackers. Commercial licensing applies only to approved source and related records within an actual accepted grant. It does not authorize sale or disclosure of other people's personal information without the required lawful basis and safeguards. Account, legal-contact and payout details are excluded from ordinary buyer or partner payloads unless necessary, lawful and separately disclosed. We share data for the purposes and recipient categories described here and in the applicable accepted terms.
- GitHub receives OAuth and GitHub App requests needed for sign-in, selected-repo listing, estimates, repository metadata, tarball download, commit timing, and admin code review.
- Neon stores account, session, GitHub App installation, profile, payout preference, submission, review, event, and archive metadata in Postgres.
- Cloudflare R2 stores encrypted submitted-codebase archive objects and object metadata.
- Vercel hosts the app and receives production analytics events and hosting/runtime metadata.
- Resend receives transactional email data when email sending is configured.
- Slack receives internal new-submission alert data if a Slack webhook is configured.
- Payment, tax, banking, legal, accounting, compliance, or support providers may receive the information required to qualify, complete, report, reverse, or investigate a payout or transaction.
- Employees, contractors, advisors, affiliates, and potential buyers may receive submission metadata, review summaries, samples, excerpts, diligence materials, or code access where needed for review, buyer outreach, transaction diligence, enforcement, or support under the Terms of Service.
- Approved AI customers and independent commercialization partners, including General Context Labs, may receive securely delivered source snapshots and approved related data under an expressly accepted commercial grant. Authorized partners may create tasks and license approved material or their resulting products onward to AI customers within that grant. An independent partner is not described as merely our processor. Before delivery, the actual signed recipient/project schedule must accept the applicable purposes, confidentiality, no-re-identification, source restrictions and withdrawal requirements; this policy does not state that GCL has already agreed to any new restrictions or amend its separate signed contract.
- Authorities, courts, platforms, rights holders, or other parties may receive information when required by law or when reasonably needed to protect rights, safety, security, users, SellCodebase, or the public.
8A. Recipient safeguards and transactional contact
Before each affected delivery, Claw records the approved recipient entities and purposes, authorized personnel/providers and processing locations, authenticated access and security controls, retention/deletion periods, withdrawal handling, and incident-notification contacts and duties. Role-appropriate data-protection agreements and lawful international-transfer safeguards must be in place for the actual processing. No unchecked assumption about a recipient, a rights checkbox or a confidentiality promise clears unresolved personal-data permissions.
Claw protects nonpublic Contributions before and after commercial delivery and takes reasonable steps to enforce recipient restrictions on learning of a breach. Confirmed incidents affecting Contributions are notified without undue delay as required by law and recipient agreements. Evaluation access is confidential, purpose-limited and time-limited, with deletion obligations; evaluation alone grants no indefinite retention, production training, onward commercial reuse or surviving commercial rights.
Owner, representative and transactional contact details support rights checks, statements, payment setup, tax/compliance checks, disputes and safety notices. Those operational communications do not enroll you in marketing. Bank and tax information belong only in the verified secure payout flow, not ordinary email, submission notes or commercial source payloads. No payout setup or payment is required for an otherwise valid, cleared commercial grant to take effect.
9. Snapshot deletion and retention
Estimate previews do not create submission records or stored code archives. Once a repository is submitted, the submission record and encrypted archive remain while the submission is pending, under review, qualified, or paid unless a deletion rule applies.
If you cancel a pending or under-review submission, or withdraw a qualified or paid submission covered by a new commercial grant from future licensing, the app marks the submission cancelled, deletes the R2 archive object if one exists, clears archive encryption material from the database, and leaves a non-code submission record in your dashboard. If the R2 delete fails, cancellation fails closed and records a retryable archive deletion error instead of pretending the code is gone.
If an admin rejects a submission, the app marks the archive for deletion and attempts to delete the R2 archive object. If deletion succeeds, archive encryption material is cleared and the record is marked deleted. If deletion fails, the rejection remains recorded and the archive state is marked as failed so an admin can follow up.
Account-deletion requests remove account, payout, submission and source records subject to necessary contract, earned-payment, security and legal retention. Storage deletion failures fail closed so the database pointers needed for retry are preserved. Necessary acceptance receipts, source/exclusion manifests, commercial license and delivery evidence, and earned royalty records may be retained for rights enforcement, accounting, payment, disputes and legal duties. Deletion must not erase an unpaid earned balance; a request may require reviewed retention rather than automatic deletion of those records. Backups, provider logs, email and tax/legal records follow their applicable retention and legal requirements.
For new grants, source withdrawal or a later exclusion stops future licensing. Surviving commercial rights require Commercial Delivery as defined in section 9 of the accepted commercial terms: a bona fide end-customer license for real non-nominal consideration, the exact source manifest, and actual delivery or authorized access. Free evaluation, partner ingestion or a paper license alone does not satisfy that test. Valid Commercial Delivery rights may survive even before customer cash is collected or your share is paid; payment debt survives separately. Earlier submissions retain their actual license and privacy limits. No archive or backup may be used to reintroduce excluded or deleted source for future licensing.
10. Security
We use selected-repository GitHub App access, short-lived GitHub App installation tokens, encrypted OAuth token fields, CSRF and origin checks, secure cookies in production, signed GitHub App setup state, database-backed rate limits, encrypted submitted-codebase archives, per-archive data keys, encrypted archive data keys, server-side archive credentials, and admin-only review routes.
No system is perfectly secure. We cannot guarantee that a repository or account will never be accessed, disclosed, altered, or destroyed. You should remove secrets, credentials, customer data, personal data, regulated data, and anything else you are not allowed to disclose before submitting a repository.
11. Cookies and tracking choices
We use first-party cookies for authentication, session state, and account security. We use Vercel Web Analytics in production for traffic and product-event measurement. We do not use third-party advertising cookies or behavioral ad trackers.
You can control cookies through your browser settings, but blocking authentication cookies may prevent sign-in, repository review, and account deletion from working.
12. Your choices and rights
You can:
- disconnect or modify the SellCodebase GitHub App from GitHub's installed-app settings;
- revoke GitHub OAuth authorization from your GitHub account settings;
- cancel a pending or under-review submission, and withdraw qualified or paid new-grant submissions from future licensing where the app allows it; contact hello@sellcodebase.com for source/exclusion or privacy requests, subject to valid existing Commercial Delivery rights and earned-payment retention;
- update payout preferences and contact details in account settings;
- delete your account from settings after confirming with a fresh session and the required DELETE confirmation;
- email us to request access, correction, export, deletion, or help with privacy rights that apply in your location.
We may need to verify your identity before acting on a request. We may decline, limit, or delay a request when allowed or required by law, including for valid Commercial Delivery rights, earned payment debt and necessary supporting records, security, fraud prevention, dispute handling, tax, accounting, legal compliance, backups, or records we need to provide the Service you requested.
13. Children
The Service is not intended for children. You may use the Service only if you meet the eligibility requirements in the Terms of Service. We do not knowingly collect personal information from children through the Service.
14. Changes
We may update this Privacy Policy as the Service changes. If we make material changes, we will update the date above and provide any additional notice required by law or appropriate for the change. We will not retroactively expand an earlier submission's scope by changing this page. Material commercial scope, recipient economics or rate changes require a new affirmative versioned acceptance for identified source. Historical receipts and agreements remain intact. Nonwaivable privacy and other statutory rights remain available.
15. Contact
For privacy requests or questions, email hello@sellcodebase.com.