https://github.com/interwebologist/short-hand-skill
AI short hand skill to maximize speed and context window on local AI system for real engineers
https://github.com/interwebologist/short-hand-skill
Last synced: about 1 month ago
JSON representation
AI short hand skill to maximize speed and context window on local AI system for real engineers
- Host: GitHub
- URL: https://github.com/interwebologist/short-hand-skill
- Owner: interwebologist
- Created: 2026-05-07T05:48:59.000Z (about 2 months ago)
- Default Branch: main
- Last Pushed: 2026-05-07T06:18:02.000Z (about 2 months ago)
- Last Synced: 2026-05-07T07:39:28.933Z (about 2 months ago)
- Homepage:
- Size: 7.81 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Short-hand Skill
A skill for AI coding assistants that enables compressed, token-efficient communication while maintaining technical accuracy. Created custom to maximize speed and stay within 65k "golden" context of local coding LLM models
## Overview
The short-hand skill helps AI coding assistants communicate more efficiently by using compressed syntax, root words, and symbol-dense formatting. This reduces token usage while preserving technical precision.
## Core Principles
- **TL;DR output by default** - Get straight to the point
- **Predictable pattern**: `[thing] [action] [why]. [next].`
- **Drop filler words**: Articles, conjunctions, pleasantries when safe
- **Prefer symbols**: Use → for causality, @ for at/target, etc.
- **One line per idea** - No periods or commas unless logically required
- **Preserve meaning** - Only compress when it doesn't risk losing meaning
## Symbolic Substitutions
| Symbol | Meaning |
|--------|---------|
| `&` | and |
| `+` | also, plus, increase |
| `→` | leads to, results in |
| `@` | at, target |
| `Δ` | change |
| `w/` | with |
| `w/o` | without |
| `=` | is, equals |
| `!=` | not equal |
| `>` | more than, better than |
| `<` | less than |
| `~` | approximately |
| `?` | query, if |
| `!` | not, invalid, negative |
| `*` | multiply, emphasize |
## Compression Patterns
- Use fragments over full sentences
- Remove duplicates
- Merge related ideas
- Prefer lists and bullets
- Use codes, tags, labels
- One word or one line when sufficient
## Examples
| Normal Speech | Short-hand |
|---------------|------------|
| not valid | `!valid` |
| database is now working let's begin the next step | `db works -> nxt step` |
| approximately 5pm | `~5pm` |
| with code and test | `w/code & test` |
## When to Use
Activate when you want:
- Faster, more concise responses
- To Maximize the "smartest" 65k context of my local LLM setup
- Reduced token usage
- Technical discussions where brevity is valued
## The skill automatically disables for:
- Security warnings
- Irreversible action confirmations
- Multi-step sequences where order matters
- When compression creates ambiguity
- When user asks for clarification
- Code, commits, PRs (use normal formatting)
## Activation
- **On**: Skill is active by default when loaded
- **Off**: Say "stop short hand" or "talk normal" to revert
- **Auto-resume**: Skill resumes after clarity-critical sections
## Boundaries
- Code/commits/PRs: Always write in normal format
- Persists until changed or session ends
- Use "stop short hand" or "talk normal" to disable manually
## Best Practices
1. Ask for clarification if output is too compressed, the AI will expand when needed for clarity