All posts
Insights

What Client System Access an Implementation Consultancy Should Grant a Staff Augmentation Engineer

Published on 9 Nov 2026

what-client-system-access-an-implementation-consultancy-should-grant-a-staff-augmentation-engineer

An implementation consultancy should give a staff augmentation engineer only the access needed for current work. On day one that means a named account in the source repository, the development and sandbox environments, and project ticketing and chat. It should not include standing access to production data or the end client's admin consoles.

Production access should be requested per task and approved by the consultancy, with each session time-boxed and logged. The stakes are contractual: the consultancy signed the end client's security terms, so every account it issues is its own liability. 724SOFTWARE writes from the supplier side of this arrangement. It places Vietnam-based engineers inside consultancy delivery teams, under onboarding processes aligned with ISO 27001:2022.

TL;DR

  • Least privilege by default: named accounts, with dev and sandbox access only on day one.

  • Production is just in time access: granted per task, approved, and set to expire automatically.

  • Role based access control should follow engagement roles in an identity provider that you or the end client control. Never use shared logins.

  • Run a vendor security assessment on the augmentation partner before seat one, with a focus on offboarding speed and AI tool policy.

  • Staff augmentation vs outsourcing: augmentation keeps access governance with you, so you carry the responsibility for it.

About the Author: This guide comes from the 724SOFTWARE delivery team. The team comprises 200+ professionals, 58% at senior level, supplying dedicated engineers to consultancies and implementation partners across 10+ countries. It is GDPR compliant and aligned with ISO 27001:2022.

Why does system access matter more than the engineer's CV?

For a consultancy adding capacity, the access decision carries more risk than the skills decision, because access is where contract liability sits. Staff augmentation adds external engineers to complement your existing team. Granting external personnel access to sensitive company information creates security risks that need deliberate management.

  • What is already happening: Verizon's 2026 Data Breach Investigations Report found that 48% of breaches involved a third party.

  • The chain is longer than it looks. It runs from the end client to the consultancy, then to the augmentation partner, then to the engineer.

  • You are the third party. In the end client's third party risk management program, the consultancy is the vendor. The augmented engineer is effectively a fourth party the client may never have assessed.

Author's opinion: "Do we trust this engineer?" is the wrong question. The useful one is "what could this account do if its credentials leaked tomorrow?"

What access should a staff augmentation engineer get on day one?

That question has a practical answer. Day-one access covers what the engineer needs to ship a first ticket, and nothing more.

System

Day one

On request, time-boxed

Default no

 

Source repository

Named account, branches, pull requests

Merge to protected branches

Org admin

Dev / sandbox

Full

n/a

n/a

Staging / UAT

Read and deploy

Configuration changes

n/a

Production application

None

Read-only logs, scoped fix windows

Standing admin

Production database

None

Masked data or read-only replica

Raw personal data export

End client's platform admin (e.g. their Dynamics 365 or Odoo tenant)

None

Scoped customisation role

Global admin

Secrets

Vault, non-production only

Production secrets per task

Plaintext sharing

Ticketing, chat, docs

Project spaces only

n/a

Whole-tenant access

What to do with it: Put this table into the statement of work for each seat. That way scope is agreed before any credentials exist.

How should role based access control be set up for augmented engineers?

The table only holds if permissions are attached to roles rather than individuals. Role based access control (RBAC) assigns permissions to a role and then assigns people to that role, so access follows the job. This structure keeps a ramp from one engineer to five manageable.

  • Create engagement-specific roles such as "ProjectX-Developer". Do not reuse internal staff roles that carry years of accumulated permissions.

  • Give each engineer one named account. Shared logins destroy the audit trail.

  • Hold accounts in your identity provider or the end client's, so you can disable them without waiting on the partner.

  • Enforce MFA and conditional access, such as managed devices only.

  • Review role membership at every sprint boundary and every headcount change.

Regulated end clients raise the bar. ISO 27001 (Annex A controls) and NIST SP 800-53 (Access Control family) define best practices for granting and governing third-party consultant access.

When should just in time access replace standing access?

RBAC sets what a role may do. Just in time access decides when. JIT grants privileged permissions for a defined task and time window, then removes them automatically.

Think of a hotel key card coded for one room until checkout. Nobody at the front desk has to remember to cancel it, because the expiry does that work. Removal becomes the default rather than a task someone has to remember. Standing production access tends to fail in exactly the moment a handover is rushed and nobody remembers to remove it.

A workable JIT workflow:

1. The engineer requests access and cites a ticket ID.

2. The consultancy's technical lead approves the scope and the window.

3. Access is granted, and the session is logged.

4. Access expires automatically.

5. The lead reviews the log at the next stand-up.

Use JIT for production fixes, data migration cutovers, database debugging and changes to the end client's admin console. NIST SP 800-207 (Zero Trust Architecture) and the CISA Zero Trust Maturity Model are recognised best-practice frameworks for managing consultant access through continuous verification.

What should a vendor security assessment of an augmentation partner check?

Your controls only work if the partner runs matching ones on its side. That is why the vendor security assessment belongs before seat one.

  • Standards evidence: Ask which ISO 27001 Annex A controls the partner operates and request evidence. A logo on a website is not evidence.

  • Employment model: Find out whether engineers are permanent staff working from a physical office or contractors sourced per request. This determines who is bound by NDAs and a disciplinary process.

  • Devices: Engineers should use managed, encrypted laptops, and client code should not reach personal machines.

  • Offboarding speed: Ask how quickly the partner revokes its side of access after a departure or a suspected incident. Get a number.

  • AI tool policy: Ask which AI coding assistants engineers use, on which accounts, and whether client code may go into them. The answer should be configurable per client.

  • Data handling: Confirm GDPR compliance and where data is processed.

Author's opinion: Offboarding speed and AI tool policy deserve the most time in the assessment, because both decide what happens to client code after the work ends.

How does staff augmentation vs outsourcing change who owns access?

The assessment carries extra weight in augmentation because of where the work happens. In staff augmentation vs outsourcing, the main operational difference is the working environment.

Staff augmentation

Project outsourcing

 

Where engineers work

Inside your team, alongside your own engineers

Usually the vendor's environment

Who grants and reviews access

You

Mostly the vendor, under contract

Main access risk

Over-provisioning, slow offboarding

Less visibility into how work is built

Augmentation gives the consultancy direct control. That control also makes the consultancy accountable for every permission it grants.

Frequently Asked Questions

Should an augmented engineer ever hold production admin rights?

Only through just in time access: per task, approved, and set to expire.

Do augmented engineers need real customer data?

Rarely. Use masked or synthetic data unless a specific ticket requires otherwise.

Who should own the engineer's accounts?

The consultancy or the end client should own them, so either can disable access instantly.

How often should access be reviewed?

Review at every sprint boundary, every role change and every ramp-up or ramp-down.

What happens when the engineer rolls off?

Revoke access on both sides the same day and rotate any secrets the engineer could see.

Can engineers use AI coding tools on client code?

Only if the end client permits it and the tool account is approved for that engagement.

About 724SOFTWARE

724SOFTWARE partners with consultancies and implementation partners to deliver dedicated engineering teams, embedded staff augmentation, and offshore development centres, scaling from 1 to 50+ pre-vetted engineers in 2-4 weeks. The company has delivery experience across 10+ countries, is aligned with ISO 27001:2022 and ISO 9001 standards, and is GDPR compliant.

It operates a follow-the-sun model with incident response under 10 minutes. As a selected Anthropic partner in Vietnam, it trains its engineers to use Claude Code in daily delivery, and it bills transparently on actual working hours, with a 95% client retention rate.

What should you do before the next seat starts?

Draft the day-one access table for that engagement and share it with your augmentation partner before any credentials are created. To compare it with how 724SOFTWARE onboards engineers into consultancy teams, get in touch at https://724software.com.vn/.

Share this article

Insights

Shrimpie Tran

AI Engineer

Keep Reading

Explore more from our experts.

View all

Stay ahead with our insights.

Get the latest on software design, strategy, and what's working in the field.

We respect your inbox. Unsubscribe anytime from any email.