Multi-agent workflow commands and skills for Claude Code. Orchestrate specialized agents to fix bugs, implement features, discover customer needs, and research issues — all end-to-end.
claude plugin marketplace add softvoyagers/claude-skills
claude plugin install softvoyagers-skills| Command | Phases | Description |
|---|---|---|
/softvoyagers-skills:bug-fix |
Diagnose → Fix → Validate → Ship | Multi-agent bug diagnosis and minimal fix with regression tests |
/softvoyagers-skills:new-feature |
Discovery → Architecture → Implement → Review → Ship | End-to-end feature implementation with customer-focused requirements |
/softvoyagers-skills:discover-feature |
Research → Synthesis → Report | Read-only customer needs analysis from 5 perspectives (Virtual Customer, UX Designer, Product Owner, Engineer, QA) |
/softvoyagers-skills:discover-feature-and-deliver-most-wanted |
Research → Selection → Architecture → Implement → Review → Ship | Discover customer needs then build the highest-impact feature |
/softvoyagers-skills:research-issue |
Input Parsing → Gather → Analyze → Report | Research non-trivial issues with 4 parallel agents (codebase, web, docs, internal knowledge) |
| Skill | Description |
|---|---|
use-gamma-slides |
Generate slide presentations using the Gamma API — prompt-driven creation, style customization, and structured content |
Each command uses the Orchestrator pattern:
- Claude acts as the orchestrator and never writes code directly
- Specialized agents are launched in parallel where possible (Explore for read-only research, general-purpose for code changes)
- Quality gates between phases ensure correctness before proceeding
- Final phase creates a branch, commits, pushes, and opens a PR
All commands enforce:
- Test naming:
MethodName_Scenario_ExpectedResult - Test structure: Arrange / Act / Assert
- Minimal change surface — only touch what's necessary
- Follow existing codebase patterns and conventions
- Tech-stack agnostic — auto-detects project tooling
MIT