Transfer of capability
The seven artifacts that make a handoff usable
Every artifact needs an owner and an acceptance check. A document that exists but cannot be used by the receiving team has not completed the transfer.
Ownership and access register
Client account owner
Accepted whenThe client controls source, hosting, domain, analytics and critical vendor accounts with recovery access
Handoff failureThe builder remains the only administrator or uses a personal account
System map
Technical maintainer
Accepted whenA new maintainer can trace requests, storage, integrations and trust boundaries
Handoff failureArchitecture exists only in chat history or the creator’s memory
Configuration contract
Deployment owner
Accepted whenRequired variables, scopes and environment differences are documented without exposing secret values
Handoff failureProduction relies on unnamed dashboard settings
Release and rollback runbook
Release owner
Accepted whenThe client can deploy a safe change and restore the last known version
Handoff failureOnly automatic deployment is understood; rollback has never been tried
Design and content source
Product or brand owner
Accepted whenFonts, assets, licenses, tokens and editable source are available and linked to the implementation
Handoff failureThe client receives compressed assets or unlicensed placeholders
Quality evidence packet
Delivery lead
Accepted whenAcceptance journeys, accessibility, performance and security scope are recorded with known limitations
Handoff failureApproval depends on a visual demo of the happy path
Support boundary
Commercial owner
Accepted whenWarranty, response path, maintenance scope and end date are explicit
Handoff failureBoth parties assume the other will handle future incidents
AI-assisted delivery needs ordinary accountability
Do not hand the client a production mystery
Prompts can be useful context, but they do not replace requirements, design decisions or a current operating model. Generated explanations can also sound complete while describing code that no longer exists. The receiving team should verify every critical artifact against the actual repository and deployment.
Record material third-party code, assets and licenses regardless of how they entered the project. If production origin matters contractually, define the required evidence before work starts; a public detector cannot reconstruct it at handoff.
- Transfer accounts before the final invoice milestone
- Use organization-owned identities and recovery methods
- Remove demo data and personal credentials
- Name known limitations in the acceptance record
Repeatable process
Run a 60-minute acceptance session
The client performs the work while the builder observes. This exposes missing capability more reliably than another presentation.
- 01
Recover access
The client signs in through its own identities and confirms account recovery for source, hosting and domain.
- 02
Trace one journey
Follow a real user action through UI, API, storage and third parties.
- 03
Change and preview
Make one harmless content or configuration change and inspect it outside production.
- 04
Release and observe
Deploy the approved change and locate logs or monitoring for the journey.
- 05
Recover
Revert or roll back the exercise and verify expected behavior.
Applied example
Handoff failure: the site works until the domain expires
A freelancer delivers a live site and repository. Six months later the domain renewal and transactional email provider are still attached to the freelancer’s personal accounts, and the client cannot recover them.
- The code delivery was complete but operational ownership was not.
- A repository cannot recover domain control or vendor billing access.
- An ownership register tested during acceptance would have exposed the dependency.
- The handoff should not be accepted until the client controls the accounts and recovery path.
Plain answers
Handoff questions
Should clients receive the prompts?+
Provide material working notes when they help maintain the product or when the contract requires them. Prompts do not replace source, decisions, licenses, tests or operational documentation.
Who should own the Vercel or hosting project?+
The arrangement should match the contract, but a client purchasing an independently operated product usually needs organization-controlled ownership and recovery—not permanent dependence on a personal builder account.
Can a README be the complete handoff?+
It can be the entry point. Acceptance still needs working access, deploy and rollback practice, system ownership and evidence that the instructions match reality.