@stillrivercode/agentic-workflow-template/AI.md

NPM package to create AI-powered GitHub workflow automation projects

# AI Assistant Context

This file provides guidance for AI assistants working with the Information Dense
Keywords Dictionary (IDK) in this agentic workflow template.

## Quick Reference

Access IDK commands via: `npx @stillrivercode/information-dense-keywords`

## Core Commands

Basic operations that form the foundation of the command vocabulary:

- **SELECT** - Find, retrieve, or explain information from codebase
- **CREATE** - Generate new code, files, or project assets
- **DELETE** - Remove code, files, or project assets (with confirmation)
- **FIX** - Debug and correct errors in code

## Development Commands

Commands for code analysis, debugging, and optimization:

- **analyze this** - Examine code/architecture for patterns and issues
- **debug this** - Investigate issues and provide root cause solutions
- **optimize this** - Improve performance and efficiency

## Documentation Commands

Commands for creating and maintaining project documentation:

- **document this** - Create comprehensive documentation with examples
- **explain this** - Provide clear, structured explanations
- **research this** - Investigate and gather comprehensive information

## Quality Assurance Commands

Commands for testing and code review:

- **test this** - Generate comprehensive test suites
- **review this** - Perform thorough code reviews

## Workflow Commands

Commands for project planning and specification:

- **plan this** - Break down complex tasks into implementation plans
- **spec this** - Create detailed technical specifications

## Git Operations

Commands for version control and GitHub interactions:

- **gh** - GitHub CLI namespace for GitHub operations
- **commit** - Create git commits with well-formatted messages
- **push** - Push changes to remote repository
- **pr** - Pull request operations (shorthand for `gh pr`)
- **comment** - Add comments to GitHub issues or pull requests

## Command Chaining

Commands can be chained together to create complex workflows:

**Sequential Chain**:
`analyze this authentication system then spec this improved version then plan this implementation`

**Parallel Operations**:
`test this user service and document this API endpoint`

**Complex Workflow**:
`debug this performance issue then optimize this query then test this solution and document this change`

## Expected Output Formats

Each command has specific expected output formats. Key principles:

### Analysis Commands

- Provide structured reports with sections for findings, recommendations, and next steps
- Include severity levels for identified issues
- Reference best practices and industry standards

### Planning Commands

- Break down into clear phases with dependencies
- Include realistic timelines and effort estimates
- Specify deliverables and success criteria

### Specification Commands

- Include functional and non-functional requirements
- Provide technical design details
- Specify testing and deployment considerations

## Integration with Agentic Workflows

This template provides:

- **AI Tasks**: Issues labeled 'ai-task' trigger automated implementation
- **OpenRouter**: All AI operations use OpenRouter API (not direct CLI)
- **Multi-model Support**: Claude, Gemini, GPT-4, Llama via OpenRouter
- **Cost Controls**: Built-in API usage monitoring and circuit breakers

## Best Practices

1. **Use Precise Keywords**: Follow IDK vocabulary for consistency
2. **Support Command Chaining**: Recognize "then" and "and" connectors
3. **Provide Structured Output**: Follow expected formats for each command
4. **Consider Context**: Leverage project-specific information and constraints
5. **Focus on Actions**: Prioritize actionable insights and recommendations

## Local IDK Dictionary

**IMPORTANT**: The `docs/dictionary/` files are for **documentation and offline
reference only**. They are NOT runtime dependencies. The actual command vocabulary
and execution logic is provided by the `@stillrivercode/information-dense-keywords`
npm package accessed via `npx`.

The local dictionary serves to:

- Provide offline reference for command definitions
- Document expected output formats
- Guide consistent AI behavior
- Enable project-specific command documentation

## Related Documentation

- [CLAUDE.md](CLAUDE.md) - Claude-specific context and instructions
- [GEMINI.md](GEMINI.md) - Gemini-specific context and instructions
- [docs/ai-workflow-guide.md](docs/ai-workflow-guide.md) - Complete AI development workflow guide
- [docs/information-dense-keywords.md](docs/information-dense-keywords.md) - Local IDK
  dictionary index (documentation only)
- [IDK Commands](https://github.com/stillrivercode/idk) - Full command vocabulary reference