Overview
Nexus coordinates multiple skills to solve complex problems in a structured and efficient way.
It does NOT simulate multiple agents.
Instead, it:
- Breaks problems into steps
- Selects the right skills
- Guides execution in order
- Ensures quality before final output
1. Core Behavior
When a task is complex:
- Understand the goal
- Break it into smaller steps
- Assign the correct skill to each step
- Execute step-by-step
- Validate results before continuing
2. Minimum Viable Workflow
Always use the simplest possible workflow.
Rules:
- Start with one skill
- Add more only if necessary
- Avoid overengineering
- Keep steps clear and linear
3. Task Decomposition
Break tasks when:
- They involve multiple concerns (UI + backend + DB)
- They touch multiple files
- They require multiple steps
Output format:
- Goal
- Steps
- Skills used per step
4. Skill Routing
Use this mapping:
- Bugs → debug + testing
- UI → frontend-design
- Database → database
- Security → security
- Features → orchestrator + thinking
5. Execution Rules
- Execute one step at a time
- Do not jump ahead
- Validate each step before continuing
- Keep changes minimal and controlled
6. Validation
Before finishing:
- Check if goal is achieved
- Verify no regressions
- Ensure code is production-ready
7. Output Structure
Always respond with:
Plan
- Goal
- Steps
- Skills used
Execution
- Step-by-step results
Validation
- What was verified
8. Anti-Patterns
Avoid:
- Doing everything at once
- Mixing unrelated concerns
- Overcomplicated workflows
- Skipping validation
Expected Behavior
This skill is working correctly if:
- Tasks are broken into clear steps
- The correct skills are used
- Output is structured and easy to follow
- Complexity is minimized
Example Usage
User request: "Fix login and improve UI"
With Nexus:
Plan:
- Fix bug → debug
- Improve UI → frontend-design
Execution:
- Step 1 → fix bug
- Step 2 → improve UI
Validation:
- Bug fixed
- UI improved