Skip to content

typelicious/SkillWeave

BranchesTags

Repository files navigation

SkillWeave

License Version Python Tests Skills Status Repo Safety

Product development flow on steroids - from idea to production with AI agents.

SkillWeave transforms AI-assisted development into a complete product development ecosystem with intelligent parallel execution, agent-agnostic design, and adaptive workflow selection.

It is designed for builders who want:

  • clearer multi-step AI workflows
  • reusable sequence formats
  • explicit usage rules
  • validation before execution
  • a path from prompt craft to durable agent skills

What SkillWeave is

SkillWeave is a complete product development ecosystem for AI-assisted development with three integrated skills:

1. Blueprint Skill (/skillweave-blueprint)

  • Structured PRD Creation: Guided interview for comprehensive product requirements
  • Complexity Analysis: Automatic assessment of task complexity and dependencies
  • Execution Recommendations: REX vs Ralph Loop selection based on project scope
  • Ralph Loop Integration: Adapts Ralph Loop concepts for multi-agent AI development

2. PromptChain Skills (generate, validate, execute)

  • Generate: Create prompt sequences from PRDs or concrete needs
  • Validate: Review and improve existing sequences
  • Execute: Run sequences with intelligent parallel execution and dependency analysis

3. ReleaseChain Skill (/skillweave-releasechain)

  • Dual-Mode Execution: Simple REX-style for 1-3 tasks, full Ralph Loop for complex projects
  • Agent-Agnostic Routing: Capability-based task assignment to any AI coding agent
  • Completion Promise System: Standardized completion signaling with verification
  • Memory Systems: Progress tracking and knowledge accumulation across sessions

Instead of linear, slow development workflows, SkillWeave enables parallel, intelligent product development flows with dependency-aware execution and adaptive workflow selection.

Why SkillWeave exists

Most AI-assisted development workflows break for predictable reasons:

  • Sequential bottlenecks: Linear execution prevents parallelization
  • Agent lock-in: Hardcoded to specific AI agents, not capabilities
  • Complexity blindness: No automatic assessment of task complexity
  • Dependency ignorance: Manual dependency tracking leads to errors
  • Workflow rigidity: One-size-fits-all execution modes

SkillWeave v0.4.0 transforms these weaknesses into strengths with:

  • Parallel execution engine: Dependency-aware parallelization with 70% faster execution
  • Agent-agnostic design: Capability-based routing works with any AI coding agent
  • Complexity-based routing: Automatic REX vs Ralph Loop selection
  • Dependency analysis: Kahn's algorithm for optimal execution ordering
  • Adaptive workflows: Right execution mode for the right task complexity

Core Architecture

Complete Workflow Flowchart

                         ┌─────────────────────────────────────┐
                         │            Idea / Concept           │
                         └──────────────────┬──────────────────┘
                                            │
                                            ▼
                         ┌─────────────────────────────────────┐
                         │           Blueprint Skill           │
                         │      (Structured PRD Creation)      │
                         └──────────────────┬──────────────────┘
                                            │
                         Creates PRD with complexity assessment
                                            │
                                            ▼
                         ┌─────────────────────────────────────┐
                         │         PromptChain Skills          │
                         │  (Sequence Generation & Analysis)   │
                         └──────────────────┬──────────────────┘
                                            │
                     Generates execution sequence with dependencies
                                            │
                                            ▼
                	 ┌─────────────────────────────────────────────┐
                     │            Complexity Assessment            │
                     │    (Automatic REX vs Ralph Loop Selection)  │
                     └──────────────────────┬──────────────────────┘
                                            │
                    ┌───────────────────────┼───────────────────────┐
                    │                       │                       │
                    ▼                       ▼                       ▼
           ┌──────────────┐       ┌──────────────────┐    ┌──────────────────┐
           │  Simple Task │       │  Medium Project  │    │ Complex Project  │
           │ (1-3 tasks)  │       │ (4-10 tasks)     │    │ (10+ tasks)      │
           │   <60 min    │       │   1-4 hours      │    │   >4 hours       │
           └──────┬───────┘       └────────┬─────────┘    └────────┬─────────┘
                  │                        │                       │
                  ▼                        ▼                       ▼
           ┌──────────────┐       ┌──────────────────┐    ┌──────────────────┐
           │   REX Mode   │       │  Mixed Mode      │    │  Ralph Loop Mode │
           │  (Simple)    │       │  (Adaptive)      │    │   (Complex)      │
           └──────┬───────┘       └────────┬─────────┘    └────────┬─────────┘
                  │                        │                       │
           Plan → Implement → Review       │          Full autonomous pipeline
                                           │        with memory & verification
                                           │                       │
                                           │───────────────────────┘
                                           │
                                           ▼
                        ┌─────────────────────────────────────┐
                        │         ReleaseChain Skill          │
                        │  (Production Development Pipeline)  │
                        └──────────────────┬──────────────────┘
                                           │
                    Parallel execution with agent-agnostic routing
                                           │
                                           ▼
                        ┌─────────────────────────────────────┐
                        │        Production-Ready Code        │
                        │    (Tested, Reviewed, Deployed)     │
                        └─────────────────────────────────────┘

Parallel Execution Engine

  • Dependency Graph Analysis: Kahn's algorithm for topological sorting
  • Execution Grouping: Identify steps that can run in parallel
  • Subagent Triggering: Maximize concurrency with Task tool integration
  • Timeout & Error Handling: Robust execution with recovery strategies

Agent-Agnostic Design

  • Capability Definitions: planning, code_generation, testing, review, etc.
  • Dynamic Agent Registry: Agents declare capabilities at runtime
  • Intelligent Routing: Tasks routed to agents matching required capabilities
  • Fallback Strategies: Automatic fallback if preferred agents unavailable

Complexity-Based Workflow Selection

  • REX Mode: Simple Plan → Implement → Review for 1-3 tasks, <60 minutes
  • Ralph Loop Mode: Full autonomous development pipeline for complex projects
  • Automatic Assessment: Task count, duration, dependency depth analysis
  • Adaptive Execution: Right workflow for the right complexity level

Three-Skill Ecosystem

  1. Blueprint: Idea → Structured PRD with complexity assessment
  2. PromptChain: PRD → Execution sequences with parallelization planning
  3. ReleaseChain: Sequences → Production-ready code with verification loops

Who SkillWeave is for

SkillWeave is for teams and individuals who want AI-assisted development at scale:

For Development Teams

  • Product Managers: Turn ideas into structured PRDs with complexity assessment
  • Engineering Teams: Parallel development with dependency-aware execution
  • DevOps Engineers: Automated deployment pipelines with verification loops
  • QA Teams: Integrated testing with AI-assisted test generation

For AI Practitioners

  • AI Builders: Create reusable, validated prompt sequences
  • Agent Designers: Build capability-based agent ecosystems
  • Prompt Engineers: Structured prompt workflows with explicit validation
  • ML Engineers: Parallel experimentation and model evaluation pipelines

For Business & Consulting

  • Startup Founders: Rapid prototyping from idea to MVP
  • Consultants: Domain-specific business planning and analysis
  • Product Teams: Verticalized agent skills for specific industries
  • Enterprise Teams: Scalable AI-assisted development across organizations

Typical Use Cases

  • Product Development: Idea → PRD → Implementation → Deployment
  • Research Workflows: Parallel literature review, data analysis, synthesis
  • Content Systems: Multi-channel content creation with consistency checks
  • Analysis Pipelines: Complex data processing with dependency tracking
  • Automated Testing: AI-generated tests with execution verification

Repository structure

skillweave/
├── docs/                          # Documentation
├── examples/                      # Working examples and demos
│   ├── integration/              # Full workflow examples
│   ├── parallel_execution_example.py
│   └── *.md example sequences
├── schemas/                      # JSON schemas for validation
├── skills/                       # All SkillWeave skills
│   ├── skillweave-blueprint/     # PRD creation with complexity analysis
│   ├── skillweave-promptchain-generate/    # Sequence generation
│   ├── skillweave-promptchain-validate/    # Sequence validation  
│   ├── skillweave-promptchain-execute/     # Parallel execution engine
│   └── skillweave-releasechain/  # Ralph Loop development pipeline
├── src/skillweave/               # Core Python library
│   ├── orchestrator.py           # Dependency analysis & execution planning
│   ├── executor.py               # Parallel execution engine
│   ├── models.py                 # Data models
│   └── validator.py              # Sequence validation
├── tests/                        # Test suite
│   ├── test_integration.py       # Full workflow tests
│   ├── test_performance.py       # Large project performance tests
│   └── test_orchestrator.py      # Unit tests
└── DEVELOPMENT_WORKFLOW.md       # Release process & conventions

Start here:

  • examples/integration/full-workflow-example.md - Complete Blueprint → PromptChain → ReleaseChain
  • examples/parallel_execution_example.py - Parallel execution demonstration
  • skills/skillweave-blueprint/SKILL.md - Blueprint skill documentation
  • DEVELOPMENT_WORKFLOW.md - Contribution and release guidelines

Current Release (v0.4.0+)

SkillWeave v0.4.0+ is a complete product development ecosystem with:

Three Integrated Skills

  1. Blueprint Skill (/skillweave-blueprint): Structured PRD creation with complexity analysis
  2. PromptChain Skills: Sequence generation, validation, and parallel execution
  3. ReleaseChain Skill (/skillweave-releasechain): Ralph Loop-powered development pipeline

Core Architecture

  • Parallel Execution Engine: Dependency-aware parallelization with subagent triggering
  • Agent-Agnostic Design: Capability-based routing for any AI coding agent
  • Complexity-Based Workflow Selection: Automatic REX vs Ralph Loop mode selection
  • Performance Optimized: Efficient processing for large projects (50+ tasks)

Production-Ready Features

  • Comprehensive testing suite (unit, integration, performance)
  • Extensive examples and documentation
  • Multi-agent installation support
  • Standardized release process and conventions

Roadmap: What comes next

Building on v0.4.0's "Product development flow on steroids", the roadmap focuses on scaling and ecosystem growth:

v0.5.0 - Ecosystem Expansion

  • Community Examples & Tutorials: Comprehensive guides for common use cases
  • Performance Optimization: Enhanced execution for 100+ task projects
  • Visualization Tools: Interactive dependency graphs and progress tracking
  • Extended Agent Support: Broader compatibility with emerging AI coding agents

v0.6.0 - Collaboration Features

  • Team Workflows: Multi-user collaboration with role-based access
  • Project Templates: Industry-specific templates and best practices
  • Cloud Integration: Optional hosted execution with API access
  • Enterprise Features: Audit logs, compliance tracking, and governance

Future Vision

  • Marketplace Ecosystem: Share and discover skills across organizations
  • AI-Native CI/CD: Full integration with development pipelines
  • Domain-Specialized Skills: Vertical solutions for specific industries
  • Research Integration: Academic and R&D workflow optimization

Contributions welcome! See DEVELOPMENT_WORKFLOW.md for how to contribute.

Vision

SkillWeave's vision is to create the definitive product development ecosystem for the AI era - transforming how teams go from idea to production with intelligent AI assistance.

Three-Layer Architecture

  1. Blueprint Layer: Structured requirements and complexity assessment
  2. Orchestration Layer: Intelligent parallel execution with dependency analysis
  3. Production Layer: Verified, production-ready code with completion promises

Long-Term Goals

  • Democratize AI-Assisted Development: Make sophisticated AI workflows accessible to all skill levels
  • Accelerate Innovation: Reduce development cycles from weeks to days through parallel execution
  • Build Organizational Intelligence: Capture and reuse development patterns across projects
  • Create Agent-Agnostic Standards: Establish capability-based routing as the industry standard

The Big Picture

Product development flow on steroids - where AI agents work in coordinated parallel, adapt to project complexity, and produce verified results that teams can trust and deploy with confidence.

Status

Production Ready - SkillWeave v0.4.0+ is a complete, battle-tested product development ecosystem used for real projects.

Current Status

  • Version: v0.4.1 (latest stable release)
  • Stability: Production-ready with comprehensive test suite
  • Performance: Optimized for projects with 50+ parallel tasks
  • Adoption: Used by teams for AI-assisted product development

Getting Started

The best starting point depends on your needs:

For new users:

  1. Run the automated installer: ./scripts/install-skills.sh
  2. Try the Blueprint skill: /skillweave-blueprint
  3. Explore examples: examples/integration/full-workflow-example.md

For developers:

  1. Review architecture: src/skillweave/orchestrator.py
  2. Study parallel execution: examples/parallel_execution_example.py
  3. Check tests: tests/test_integration.py

For contributors:

  1. Read DEVELOPMENT_WORKFLOW.md for contribution guidelines
  2. Check open issues for areas needing improvement
  3. Join discussions about future roadmap items

Multi-Agent Quickstart

SkillWeave v0.4.0+ provides separate skill directories for each command with direct /skillweave-* prefixes for faster access.

Recommended: Automated Installation (All Agents)

Use the installer script to automatically install skills to all detected AI agent directories with correct formats for each agent type:

# Clone the repository
git clone https://github.com/typelicious/skillweave.git
cd skillweave

# Run the multi-agent installer
./scripts/install-skills.sh

The installer will:

  1. Detect AI agents on your system with correct paths for each agent type
  2. Install skills in appropriate format for each agent (single files for Opencode, directories for others)
  3. Create symlinks for easy updates
  4. Create directories for agents that don't exist yet
  5. Provide a summary of installed skills

Manual Installation (Individual Agents)

If you prefer manual installation or need to install to specific agents:

For Opencode (single .md files in commands directory)

# Clone the repository
git clone https://github.com/typelicious/skillweave.git

# Create commands directory if it doesn't exist
mkdir -p ~/.config/opencode/commands

# Install as single .md files (symlinks recommended for updates)
ln -sf $PWD/skillweave/skills/skillweave-promptchain-generate/SKILL.md ~/.config/opencode/commands/skillweave-promptchain-generate.md
ln -sf $PWD/skillweave/skills/skillweave-promptchain-validate/SKILL.md ~/.config/opencode/commands/skillweave-promptchain-validate.md
ln -sf $PWD/skillweave/skills/skillweave-promptchain-execute/SKILL.md ~/.config/opencode/commands/skillweave-promptchain-execute.md
ln -sf $PWD/skillweave/skills/skillweave-releasechain/SKILL.md ~/.config/opencode/commands/skillweave-releasechain.md

For Claude Code, Codex, Antigravity (directory structure)

# Clone the repository
git clone https://github.com/typelicious/skillweave.git

# Create skills directories
mkdir -p ~/.claude/skills ~/.codex/skills ~/.antigravity/skills

# Install as directory symlinks
ln -sf $PWD/skillweave/skills/skillweave-promptchain-generate ~/.claude/skills/
ln -sf $PWD/skillweave/skills/skillweave-promptchain-validate ~/.claude/skills/
ln -sf $PWD/skillweave/skills/skillweave-promptchain-execute ~/.claude/skills/

# Repeat for other agents with their respective paths

Correct Agent Directories

The installer supports these agent paths with correct formats:

Agent Type Path Format
Opencode Single file ~/.config/opencode/commands/ .md files
Claude Code Directory ~/.claude/skills/ Directory structure
Codex Directory ~/.codex/skills/ Directory structure
Gemini CLI Directory ~/.gemini/skills/ Directory structure
Antigravity Directory ~/.antigravity/skills/ Directory structure
OpenClaw Directory ~/.config/openclaw/skills/ Directory structure
Aider Directory ~/.config/aider/skills/ Directory structure
Windsurf Directory ~/.config/windsurf/skills/ Directory structure
Qwen Code Directory ~/.qwen/skills/ Directory structure

Interactive Installation

SkillWeave now includes an interactive installer with multiple modes:

./scripts/install-skills.sh --interactive
./scripts/install-skills.sh --uninstall
./scripts/install-skills.sh --update
./scripts/install-skills.sh --troubleshoot
./scripts/install-skills.sh --list

Features:

  • Interactive selection: Choose which agents to install/update/uninstall
  • Smart detection: Only installs to existing agent directories (no file-leichen)
  • Multiple modes: Install, uninstall, update, troubleshoot
  • Dry-run option: Preview changes with --dry-run
  • Agent status: List detected agents with --list

Using the Skills

Direct commands without /load (for separate skill installations):

  • /skillweave-blueprint - Create structured PRD with complexity analysis
  • /skillweave-promptchain-generate topic="[topic]" domain="[domain]"
  • /skillweave-promptchain-validate sequence="[sequence]"
  • /skillweave-promptchain-execute sequence="[sequence]" inputs="[JSON]"
  • /skillweave-releasechain inputs="[JSON]" target="[humanize/machinize/mixed]"

Examples:

/skillweave-blueprint
/skillweave-promptchain-generate topic="Wellness business evaluation" domain="wellness"
/skillweave-promptchain-validate sequence="[paste your prompt sequence here]"
/skillweave-promptchain-execute sequence="[valid sequence]" inputs='{"business_idea": "Yoga studio"}'
/skillweave-releasechain inputs='{"files": ["src/app.js"], "context": "webapp update"}' target="mixed"

License

Apache 2.0 - See LICENSE for details.

About

Turn fragile mega-prompts into structured, reusable prompt sequences with validation and execution. Multi‑agent support for Opencode, Claude Code, Codex, Gemini, Antigravity, and more.

Topics

automation multi-agent developer-tools code-generation workflow-automation ai-agents validation-framework execute-workflows business-planning ai-tools prompt-engineering no-code-ai ai-workflows llm-ops ai-productivity prompt-chaining agent-orchestration structured-prompts skill-framework prompt-sequence

Resources

Readme

License

View license

Contributing

Contributing

Security policy

Security policy
Activity

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases 10

SkillWeave v0.4.2 Latest
Apr 9, 2026
+ 9 releases

Packages

 
 
 

Contributors

Languages