# Decide

> Auto-learns your decision patterns. Grows autonomy with trust, confirms before assuming.

- Skill: `johnalbertini14-glitch/decide` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add johnalbertini14-glitch/decide`
- Raw SKILL.md: https://api.skillmd.com/api/skills/johnalbertini14-glitch/decide/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: johnalbertini14-glitch (https://skillmd.com/u/johnalbertini14-glitch)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/johnalbertini14-glitch/decide

---


## Auto-Adaptive Decision Memory

This skill auto-evolves. Observe decisions, detect patterns, confirm before internalizing.

**Core Loop:**
1. **Observe** — Notice decisions without prompting
2. **Pattern** — After 2+ consistent choices, propose confirmation
3. **Confirm** — Ask: "Should I default to X for Y situations?"
4. **Store** — Only after explicit yes, add below
5. **Evaluate** — Before applying, check if context requires re-asking

Check `confidence.md` for levels. Check `exceptions.md` for when to ask even with confirmed preferences.

---

## Scaling

Preferences grow. Don't load everything—search when relevant.

- **On task start:** Scan section headers, load only matching categories
- **If unsure:** Search keywords in stored preferences before asking user
- **Keep entries atomic:** One decision per line, easy to grep

This skill can hold hundreds of preferences without bloating context.

---

## Entry Format

One line, max 10 words: `context: preference (level) [notes]`

Examples:
- `mobile app: Flutter (confirmed)`
- `deploy: always preview first (pattern)`
- `formatting: no prettier (locked) [strong]`

---

### Stack
<!-- Tech decisions. Format: "context: choice (level)" -->

### Process
<!-- Workflow decisions -->

### Style
<!-- Aesthetic/convention decisions -->

### Comms
<!-- Communication preferences -->

### Never
<!-- Explicitly rejected approaches -->

---

*Empty sections = nothing learned yet. Observe and propose.*