Produce real, clickable browser prototypes where feasible.
I want to experience the different approaches, not merely read design descriptions.
Provide:
- A concept gallery.
- Desktop previews.
- Mobile previews.
- Interactive navigation.
- Representative editing interactions.
- Light/dark switching.
- English/Norwegian previews.
- Screenshots.
- A comparison of strengths, weaknesses and tradeoffs.
Use the approved development environment and synthetic data.
Do not deploy these prototypes to production.
If external inspiration or imagery is used, respect usage rights and attribution.
## 8. Challenge existing design decisions
Do not assume the current navigation is best.
Consider whether Jobjakt really needs its existing number of navigation destinations.
Explore grouping functionality into intuitive areas such as:
- Home.
- Explore.
- Applications.
- Career.
- Calendar.
- Account.
These are illustrative, not mandatory.
Think about whether jobs, correspondence, documents, interviews and follow-ups should be connected within one application workspace instead of spread across unrelated pages.
Look for opportunities to reduce the number of clicks needed to complete everyday tasks.
Keep advanced features discoverable.
Avoid excessive boxes inside boxes.
Use typography, spacing and purposeful layout divisions rather than decorative cards everywhere.
Do not mistake minimalism for hiding functionality.
## 9. Evaluate implementation strategies
Compare at least:
A. Thorough redesign using the existing React/component architecture.
B. Major frontend restructuring with new shared architecture and components.
C. A fresh frontend implementation built around the existing backend APIs.
Assess:
- Expected design freedom.
- Risk to preserved functionality.
- Development effort.
- Testing and migration cost.
- Accessibility.
- Performance.
- Maintainability.
- European localisation.
- Compatibility with current workstreams.
Recommend the best approach with evidence.
Do not automatically choose the most expensive rewrite.
Do not begin replacing the production frontend without owner approval.
## 10. Owner design review
Once the five concepts and comparison are ready, present them for review.
I want to be able to:
- Choose a preferred concept.
- Combine features from different concepts.
- Reject individual visual patterns.
- Request changes.
- Continue iterating until I am satisfied.
Do not assume I will choose the first recommended design.
Do not begin full replacement implementation until I explicitly approve the selected direction and implementation strategy.
While waiting for design feedback, continue independently authorised work elsewhere in the roadmap.
---
# PART C — Continue everything else
This is equally important.
**The new design exploration must NOT stop the rest of Jobjakt development.**
Continue the previously approved maximum-backlog execution directive.
Work through the entire existing TODO, including:
- Phase 10 eligible acceptance and integrity work.
- Product Review preparation.
- International Global Job Search.
- European localisation.
- EUROPE-READY geographic support.
- All applicable performance optimisations.
- Public homepage work within approved scope.
- Security and dependency remediation.
- Release readiness.
- Database and operational improvements.
- Existing technical debt.
- Outstanding functional and integration defects.
Respect existing gates.
Do not use the potential redesign as an excuse to abandon substantial previously completed work.
## Parallel execution
Where Codex supports safe parallel workers, establish independent ownership.
Suggested workstreams:
**Worker A — Existing A3 audit**
Establish actual completion and preservation status.
**Worker B — From-scratch UX research and concepts**
Produce new design options and prototypes without modifying live application behaviour.
**Worker C — European expansion**
Continue localisation, geographical architecture and independently authorised Global Jobs work.
**Worker D — Backend, security and performance**
Continue validated backend optimisations, dependency remediation, data integrity and operational improvements.
Use actual worker availability and reasonable hardware limits.
Run fewer workers if necessary.
Workers must not collide on shared files or mutable databases.
Assign separate branches/worktrees where appropriate.
## Avoid wasted implementation
Since we may replace the visual design completely, deprioritise discretionary cosmetic polishing of existing A3 components.
Still fix:
- Security vulnerabilities.
- Data loss.
- Incorrect functionality.
- Accessibility failures.
- User-isolation problems.
- High-impact defects.
Continue implementation of reusable business logic, APIs, localisation infrastructure, data models, tests and other design-independent work.
Do not remove accepted functionality because its existing presentation might be replaced.
## Continuous progress
After every validated implementation package:
1. Update the authoritative TODO.
2. Record tests and evidence.
3. Commit and publish when authorised.
4. Select the next independently eligible backlog task.
5. Continue immediately.
Do not stop merely to deliver small milestone summaries.
Maintain separate statuses for:
- Existing A3 completion.
- New redesign research.
- New redesign owner approval.
- New redesign implementation.
- Wider roadmap implementation.
- Final product acceptance.
Do not falsely mark the proposed redesign as implemented.
## Approval and safety boundaries
Maintain all existing restrictions:
- No autonomous orchestrator.
- No merge into main.
- No production deployment.
- No destructive migrations.
- No unapproved third-party costs.
- No bypassing mandatory owner decisions.
- No unqualified production AI/provider activation.
- No removal of protected historical data or functionality.
- No claims of acceptance without evidence.
Use prior verified test evidence where still applicable.
Batch owner decisions where possible.
However, the new design selection is an explicit mandatory owner gate.
Do not self-approve it.
## Required deliverables
### A3 assessment
A factual report explaining whether the existing redesign is complete, partial or awaiting acceptance.
### New design exploration
Five distinct interactive visual concepts with desktop/mobile coverage, screenshots and a comparative recommendation.
### Implementation recommendation
A reasoned comparison of updating, restructuring or replacing the frontend.
### Preservation analysis
Evidence showing how all current functionality and routes can survive a new design.
### Roadmap progress
Completed implementation, test results, commits, published changes, blockers and remaining authorised tasks.
### Owner decisions
A consolidated design-review package plus any other genuinely blocked decisions.
## Final directive
Begin immediately.
**Assess A3, explore a completely new Jobjakt design, and keep developing the remaining roadmap concurrently wherever safely possible.**
I am open to starting the user experience over from the ground up.
Do not restrict the exploration to the old A3 design.
Do not discard existing application functionality.
Do not pause other independent development work while design choices await my review.
Make substantial progress for as long as the execution environment permits.
The objective is to build the best possible Jobjakt experience for European jobseekers, not simply finish the existing design checklist.
Owner directive archived in DESIGN_REASSESSMENT_DIRECTIVE.md; SHA256 5b7fc4f837569f01286ba57fa07954d6fafae716611a3898387c595314a72de5. Exploration is released from A3 visual/navigation constraints. Five NEW distinct clickable concepts are requested; preserve original A3 prototypes and all functionality. No replacement implementation until explicit selected-direction AND strategy approval. A3 final visual acceptance remains absent. Separate statuses: existing A3 audit ACTIVE; new research/prototypes PENDING; new owner approval PENDING; replacement implementation NOT AUTHORISED; wider roadmap CONTINUES. All24 EU/all60 planned locale scope retained. Deprioritise discretionary A3 cosmetics; continue integrity/accessibility/business logic/performance and other eligible work.
Read-only/report worker design_reassessment_audit dispatched for34routes/57groups and IM01–10; owns only docs/ui-modernisation/reassessment/A3_ASSESSMENT.md. Second research worker creation failed at agent thread limit; no research/prototype work claimed. Root must perform research/prototypes sequentially or reuse an available worker without idle retries. Next: reconcile CV proposal baseline, inspect current official product interfaces, build five new concepts in an isolated design lab, capture real application desktop/tablet/mobile EN/NB light/dark separately. Existing checkpoint8d17e01c and A3 reported2c5d0451 remain baseline pending fresh audit.
## Recovery audit continuation
Roadmap encoding repair798c410e published/refverified; verified c18c69f8 historical checkpoint/status preserved verbatim with legitimate new entries.36 non-orchestrator worktrees inspected; snapshot retained at E:/Storage/Projects/Active/temp docs/jobjakt-local-build/browser-restoration-20261117/recovery-worktrees.json. Main checkout unrelated dirty files untouched. A3 remote2c5d0451 and performance remote2a212c78 verified.
Owner directive archived in DESIGN_REASSESSMENT_DIRECTIVE.md; SHA256 5b7fc4f837569f01286ba57fa07954d6fafae716611a3898387c595314a72de5. Exploration is released from A3 visual/navigation constraints. Five NEW distinct clickable concepts are requested; preserve original A3 prototypes and all functionality. No replacement implementation until explicit selected-direction AND strategy approval. A3 final visual acceptance remains absent. Separate statuses: existing A3 audit ACTIVE; new research/prototypes PENDING; new owner approval PENDING; replacement implementation NOT AUTHORISED; wider roadmap CONTINUES. All24 EU/all60 planned locale scope retained. Deprioritise discretionary A3 cosmetics; continue integrity/accessibility/business logic/performance and other eligible work.
Read-only/report worker design_reassessment_audit dispatched for34routes/57groups and IM01–10; owns only docs/ui-modernisation/reassessment/A3_ASSESSMENT.md. Second research worker creation failed at agent thread limit; no research/prototype work claimed. Root must perform research/prototypes sequentially or reuse an available worker without idle retries. Next: reconcile CV proposal baseline, inspect current official product interfaces, build five new concepts in an isolated design lab, capture real application desktop/tablet/mobile EN/NB light/dark separately. Existing checkpoint8d17e01c and A3 reported2c5d0451 remain baseline pending fresh audit.
## 18 November combined journeys and user-summary projection published
A3 `2c5d04513eb02a9b3771fde5c27ea3e378fc5d92` published/ref-readback verified. Combined browser evidence171e616b integrated213793c1: fresh Release API on combined checklist/analytics source plus unchanged qualified export;4/4 connected discovery/calendar cases,8scans/8screenshots,0violations,0skips/flaky. Prior distinct-case counts overlap; no full/native-language/visual acceptance.
- [ ] Independently audit A3:34routes/57groups,IM01–10 evidence,owner gates,current rendered desktop/tablet/mobile EN/NB light/dark; no final acceptance inferred.
- [ ] Research current public interfaces and create five distinct clickable concepts covering all12 requested screen families and connected journeys. Preserve original prototypes.
- [ ] Continue wider authorised roadmap and recovered CV journey while design exploration proceeds.
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.