Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

Employee Trial Shift

Open
  • Is something wrong?
#527 3 comments 0 reactions 1 assignee View on GitHub

Maintainers usually reply within 2 days

@agentvivekkumar is already working on this.

Since Oct 25, 2026.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
20/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Active
Tech stack
typescript
Domain
ai-infra-agents

Research direction

The issue describes an initial local foundation, including structured scenarios, evaluation logic, simulated tool denial, reports, and a testing dialog, but names no files or tests. Start by reviewing the existing agent architecture and the local foundation mentioned in the issue. The work is complete when maintainers agree on integration and isolation, and real employee execution is verified to be isolated from production resources.

Written by the indexing model from the issue text.

Description

enhancement Require Customer
Problem / motivation

Currently, business owners can create and configure AI employees, but there isn't a dedicated way to evaluate whether those employees will perform their assigned responsibilities correctly before giving them access to real business systems.

For example, a business owner might create a customer-support employee responsible for handling refunds. Before deploying it, they should be able to verify that it follows company policies, handles unusual situations appropriately, and respects its permissions.

Without a safe testing environment, owners have limited evidence of how their employees will behave.

This is particularly important for small businesses that may not have technical expertise or the resources to recover from automation mistakes.

The goal is to let business owners answer a simple question:

"Can I trust this AI employee to do its job before I let it work with real customers and business data?"

Proposed solution

I propose adding an Employee Trial Shift feature to Don't Be Michael.

After creating or configuring an AI employee, users could run a simulated trial shift using fictional business data before deploying that employee.

Proposed workflow

  1. Select an existing AI employee.
  2. Click "Test Employee."
  3. Choose a fictional business scenario relevant to the employee's responsibilities.
  4. Execute the employee in a properly isolated testing environment with simulated tools and business data.
  5. Evaluate its responses, decisions, policy compliance, and attempted actions.
  6. Generate an evidence-based report showing what passed, what failed, and what requires attention.

Architecture

The feature should be reusable across all business types rather than implementing industry-specific logic.

A shared evaluation engine would process configurable scenario definitions, simulated tool responses, expected outcomes, and behavioral checks.

The initial implementation could support customer support, recruitment, and restaurant scenarios using the same underlying framework.

Real-agent execution would require an enforced isolation boundary that prevents access to production credentials, files, networks, and live integrations. Unknown tools should be denied by default.

This proposal is distinct from #371, which focuses on AI engine trial shifts. Here, the focus is evaluating a configured employee's behavior and instructions before deployment.

Implementation progress

I have already built an initial local foundation within the existing application:

  • Structured testing scenarios
  • Policy and behavioral evaluation logic
  • Denial of unlisted simulated tools
  • Evidence-based reports
  • An initial employee testing dialog
  • 27 new passing automated tests

Type checking and the production build also pass.

Current limitation: This foundation does not yet execute real AI employees in a verified isolated environment. The existing demonstration uses scripted behavior, and I do not consider it a complete sandbox.

Feedback requested

Before proceeding with the full implementation, I'd appreciate maintainer feedback on:

  • Whether this feature fits the project's direction.
  • How real employee execution should integrate with the existing agent architecture.
  • Which isolation approach the project would prefer.

I'm happy to continue developing the feature and submit the implementation as a PR once the architecture is agreed upon.

Screenshot, mockup or recording

No response

Alternatives considered

No response

Area

Other

Pre-flight
  • I attached a screenshot, mockup or recording above.
  • I searched existing issues and discussions, and this isn't a duplicate.
Dominant language
TypeScript
Stars
273
Forks
211
Avg merge
2d 9h
Merged PRs (30d)
367

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from agentvivekkumar/dontbemichael

All issues in agentvivekkumar/dontbemichael

Similar issues

More TypeScript issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.