Hi
are there any best practice how to handle robot and developer user or any practical experience to share?
(I use ROBOT1, ROBOT2 etc. for robot user and DEV1, DEV2,...for developer personal users to illustrate below.)
Background: assume we are a team of 5 developers who work as a team on 3 automation projects in parallel and implementing 50 projects in various business areas (Finance, HR, Sales) and various systems throughout the year. A development proejct incl. UAT takes 4-6 weeks, later on we have to expect weekly need for some kind of support / bugfix / analysis. we have virtual machines (VM) as Interactive Client for each developer and virtual machines for each Robot (public resource PC)
Assumptions:
AD is suggested for BluePrism authentification in corporate environment
in order to develop you need to be logged in to the desktop of the interactive client VM and start BluePrism there
you should be logged in with a user that has credentials for accessing the applications which are mostly single-sign-on (SSO) nowadays
development for RPA also needs to be done in productive systems, we ahve realized that it will not be possible to do all in test systems due to various reasons (replicating specific data, different HTML id in PRD systems needs re-spying) - we see it more like training a new employee: you start training in a test system, but once they start working in the productive system you might want to look over the shoulders and assits to finish your training
Requirements:
traceability of transactions in our corporate systems (named accounts) - not allowed that 5 developers use the same Robot
traceability of changes in BluePrism scripts
Concerns, Questions, Ideas:
we can't keep adding authorizations to each developer user:
they would have very powerfull users which is a risk already
we do not want that they permanently have the authorization to issue credit notes
authorizations needed for different processes might be conflicting and even not allowed by our Internal Controls due to seggregation of duties need (e.g. issue and approve a credit note)
still we need to be able as developers to quickly get access (in a controlled / logged way?) to a process for support / analysis / bugfixing
if developers start using the robot like ROBOT1 during development, then we do not see in BluePrism save comments, who saved a change (it is anonymous) - and at that time they can use the client system like SAP with the robot account