[docs] Add dictation prompt instructions for speech-to-text optimization

[github-actions]

Overview

This PR adds comprehensive dictation instructions to help rephrase speech-to-text transcriptions into proper agentic workflow prompts for the gh-aw repository.

What's Included

Project-Specific Glossary (105 terms)

Extracted from the codebase analysis:

• Package names and types (workflow, parser, cli, console, logger)
• Technical components (MCP servers, engines, frontmatter, lockfile)
• Commands (compile, audit, logs, mcp, recompile)
• Configuration fields (safe-outputs, permissions, tools, cache-memory, toolset)
• GitHub concepts (workflow_dispatch, pull_request, issues, permissions)
• Repository-specific terms (agentic workflows, gh-aw, activation, execution)
• File types and formats (markdown, .lock.yml, frontmatter, YAML)
• Tool types (edit, bash, github, playwright, web-fetch, web-search)
• Operations (recompile, fmt, lint, test-unit, test-integration)

Rephrasing Guidelines

1. Fix Ambiguous Terms: Correct common misrecognitions like "ghaw" → "gh-aw", "work flow" → "workflow", "safe outputs" → "safe-outputs"
2. Sentence Structure: Transform casual dictation into clear, imperative agentic prompts
3. Tone Adjustment: Convert casual speech into professional technical instructions
4. Context Enhancement: Add implicit context that may be missing from spoken requests

Transformation Examples

8 comprehensive examples showing how to convert dictated input into proper agentic prompts:

• Workflow creation
• Compilation tasks
• Investigation requests
• Build and test sequences
• MCP configuration
• Safe outputs setup
• Debugging workflows
• Engine configuration

How It Was Generated

1. Analyzed Go source files to extract package names, types, and functions
2. Scanned workflow files for configuration patterns and frontmatter fields
3. Reviewed documentation for key terms and concepts
4. Examined constants for default values and configuration options
5. Identified common patterns in CLI commands and Make targets

Usage

This instruction file enables AI agents to:

• Recognize project-specific terminology even when mispronounced
• Convert casual speech-to-text input into actionable agentic prompts
• Apply consistent formatting and structure to dictated requests
• Add necessary context that may be implicit in spoken commands

Benefits

• Improved accuracy: Speech-to-text systems will better recognize technical terms
• Consistency: All dictated prompts follow the same professional format
• Efficiency: Reduces back-and-forth clarification by providing complete context
• Accessibility: Makes voice-based interaction with the repository more effective

Validation

✅ File created at .github/instructions/dictation.instructions.md
✅ Contains 105 project-specific terms (target: ~100)
✅ Includes comprehensive rephrasing guidelines
✅ Provides 8 example transformations
✅ Alphabetically sorted glossary for easy reference

AI generated by Dictation Prompt Generator