Outsourced ProgrammersPowered by StealthAgents

How long do revoked offshore credentials stay usable after an exit?

A protocol for measuring revocation latency, the credentials that survive an exit, and the evidence that the old login has stopped working.

How long do revoked offshore credentials stay usable after an exit?
WORK_LANE / RESEARCH BRIEF / SECURITYREVIEW QUEUEACCEPTANCEPASSowner review requiredEVIDENCE TRAILTICKET → BRANCH → REVIEW → CLIENT RELEASE
Table of contents
  1. Research context
  2. Practical application
  3. Operating controls
  4. Sources
  5. Frequently asked questions

Research context

This protocol asks how long an offshore credential remains usable after the account owner has left the engagement. The unit is one credential in scope for one exit, identified by the account identifier and the exit record. The study measures the time from the exit decision to the last successful use of the old credential. It does not judge whether the exit was planned or whether the credential was exposed before removal.

Define the exit decision time as the timestamp recorded by the client owner, not the time the contractor stopped working. A person can stop working before the formal exit, and the gap matters for the risk window. Record both times and report the difference. The decision time is the fair start of the revocation clock.

Set the scope to one client, one identity provider, and one offboarding policy. Record the credential types, the systems in scope, and the person who confirms each revocation at the start. If the policy changes during the window, keep the earlier definition for exits already in progress and analyze the later cohort separately. A new tool can look like a faster process when it is only a different queue.

Measure successful authentication attempts as well as the formal disable event. A credential is usable if a system accepts it, even when the directory says the account is disabled. Record the last successful use and the source address. If the system cannot log the source, state the limitation and use the application log instead.

Define the population as every exit in the window, including exits that were planned, urgent, or later reversed. The unit is the credential, not the person, because one person can hold several credentials with different owners. A credential created during the exit process is a new unit linked to the same exit. Test the definition on two recent exits before collection begins.

Record the credential type, the system, the owner, the last use before the exit, the revocation method, the revocation time, and the first failed authentication. Add the person who verified the result and the evidence they used. Keep the exit record and the access log linked. A second reviewer should be able to rebuild the timeline without asking the team.

Practical application

Separate a credential that was disabled from one that was also revoked at the provider. Disabling an account in the directory may not invalidate a token that the provider issued earlier. Record the provider revocation and the session invalidation separately. The two steps answer different questions and should not be collapsed into one checkbox.

Check the indirect credentials, such as personal access tokens, SSH keys, package registry tokens, and delegated mailboxes. These are often created outside the identity provider and are missed by a directory-only review. Record the token identifier and the owner who can revoke it. The study should say which credential types were not visible.

Report the number of exits, the number of credentials in scope, the median revocation time, and the share that remained usable after the exit date. Show the counts by credential type and by revocation method. Present the share with no recorded verification and the longest observed delay. Use the same denominator for every rate and show exits with incomplete records.

Separate a delay caused by a missing owner from one caused by a technical limitation. A vendor portal with no administrative API is a different problem from a system whose owner was on leave. Report the two categories separately so the client can choose the right fix. A new tool will not solve an ownership gap.

Inspect the slowest revocations and describe the pattern without naming the engineer. A cluster around a particular system may point to a shared password, while a cluster around a particular day may point to a batch process. Compare the result with the exit type and the number of credentials. Report urgent exits separately because they may use a different path.

Operating controls

Do not treat a fast disable as proof that the credential is gone. An active session or a refresh token can outlive the visible account state. Record the session invalidation and the token revocation as separate evidence. A fast directory change with no provider check is an incomplete result.

The record cannot show whether the credential was used for an unauthorized action before revocation. It can show only that the system accepted a login after the exit. A late revocation is not proof of a breach, and an on-time disable is not proof of safe access. The protocol also cannot capture a credential that was copied outside the monitored systems.

A useful additional view is the share of exits where the contractor reported a credential that the client had not inventoried. A self-report is not a control, but it can reveal a shadow integration. Record the reported credential and whether it was revoked. That field turns the exit into a small discovery exercise.

Give the result to the client security owner and the offshore lead and ask them to choose one observable change. The change might be a same-hour provider revocation, a named owner per credential type, a pre-exit inventory, or a quarterly old-credential test. Record the owner, the effective date, and the measure that will show whether the change worked. Repeat the protocol with the same definitions.

Close the review with one change and a date to re-measure. The change should address the credential type that caused the longest delay, not the one that is easiest to fix. Record the expected effect and the evidence that will prove it. A revocation program that never tests an old credential is relying on a policy, not a control.

Sources

  1. NIST SP 800-53 security and privacy controls
  2. NIST SP 800-63B digital identity guidelines
  3. CIS Benchmarks
  4. CISA Secure by Design

Frequently asked questions

What should happen first?

Begin with a bounded ticket, approved access, explicit acceptance evidence, and a named reviewer.

Who approves production changes?

The client's technical owner keeps final merge and release authority.

Related research

What share of outsourced production work uses a shared admin account?

What share of offshore delivery decisions are recorded in writing?

Build a programming lane your technical owner can review.

Bring the stack, backlog, acceptance checks, and release boundaries. We will help scope the role around the work.

Discuss your programming needs