[docs] Update dictation skill with comprehensive project glossary

[github-actions]

Summary

Updated the dictation skill file (skills/dictation/SKILL.md) with a comprehensive project glossary extracted from documentation and enhanced error correction rules.

Changes

1. Project Glossary (1000 terms)

• Extracted ~1000 project-specific technical terms from all documentation files in docs/src/content/docs/
• Terms filtered to exclude generic words and focus on gh-aw-specific terminology
• Alphabetically sorted for easy reference
• Includes: configuration keys, tool names, command names, GitHub concepts, operational patterns

2. Speech-to-Text Error Correction
Added correction rules for common misrecognitions:

• GitHub/Git terms: "get hub" → github, "pull request" → pull-request
• Workflow config: "front matter" → frontmatter, "safe outputs" → safe-outputs
• AI engines: "co-pilot" → copilot, "code x" → codex
• Commands: "G.H. A.W." → gh-aw
• Technical acronyms: "M.C.P." → MCP, "A.P.I." → API

3. Text Cleanup Instructions

• Remove filler words (um, uh, you know, like, basically, actually)
• Improve clarity (remove redundancy, use active voice)
• Maintain technical accuracy (preserve terminology, don't simplify concepts)

4. Clear Guidelines
Explicitly instructs the AI to:

• ✅ Fix speech-to-text errors and improve text quality
• ❌ NOT plan steps or provide code examples
• ❌ NOT add information not in original text

Technical Details

Term Extraction Process:

1. Scanned all markdown files in docs/src/content/docs/
2. Extracted technical terms using regex pattern matching
3. Filtered out common English words and generic terms
4. Prioritized frequently-used project-specific terminology
5. Limited to exactly 1000 terms for manageable size

Files Modified:

• skills/dictation/SKILL.md - Complete rewrite with new glossary and improved instructions

Testing

Verified the skill file:

• ✅ Contains proper SKILL.md frontmatter (name, description)
• ✅ 1000 project-specific terms extracted from documentation
• ✅ Speech-to-text correction rules for common misrecognitions
• ✅ Text cleanup and improvement instructions
• ✅ Clear guidelines to avoid planning/examples

Related

This skill is used by workflows and agents to fix speech-to-text errors in dictated content, particularly useful for:

• Fixing voice-to-text input in workflow descriptions
• Cleaning up dictated issue/PR comments
• Improving clarity of dictated documentation

Closes: N/A (enhancement)

References:

• §22030917563

AI generated by Dictation Prompt Generator

• expires on Feb 17, 2026, 6:21 AM UTC

---

---

✨ PR Review Safe Output Test - Run 22035866590

💥 [THE END] — Illustrated by Smoke Claude

[github-actions]

SEC-004: Add content sanitization to safe-output handlers
[docs] Update sandbox documentation to reflect SRT removal
GitHub MCP: ✅ | Serena MCP: ✅ | Playwright: ✅
File write+cat: ✅ | Build: ✅
Overall: PASS

AI generated by Smoke Codex

[github-actions]

Smoke Test Results

PRs Tested:

• [docs] Update dictation skill with comprehensive project glossary #15888: [docs] Update dictation skill with comprehensive project glossary
• Simplify sandbox/firewall detection - consolidate into isSandboxEnabled #15859: Simplify sandbox/firewall detection

Results:

• ✅ GitHub MCP
• ✅ Safe Inputs GH CLI
• ⚠️ Serena MCP (skipped)
• ✅ Playwright
• ✅ File Writing
• ✅ Bash Tools
• ✅ Discussion Interaction
• ✅ Build gh-aw
• ✅ Discussion Creation
• ✅ Workflow Dispatch
• ✅ PR Review

Overall Status: PASS

@app/github-actions

AI generated by Smoke Copilot