# Self Improvement

> Stores user corrections and preferences as learnings to improve future responses. Use when the user corrects a mistake, states a preference, or says 'remember that', 'from now on', or 'I prefer'.

- Skill: `jason-easyazz/self-improvement` (Agent Skill)
- Install (CLI): `npx skillmds@latest add jason-easyazz/self-improvement`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jason-easyazz/self-improvement/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools, DevOps & Infra
- Author: jason-easyazz (https://skillmd.com/u/jason-easyazz)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/jason-easyazz/self-improvement

---

# Self-Improvement

## When to Use

Runs in the background after every conversation. Activates explicitly when the user corrects Zoe, states a preference, or uses trigger phrases like "remember that", "from now on", "I prefer", or "don't do that again".

## Workflow

1. **Detect correction signal** — scan the user's message for corrections ("actually, I meant…"), preference statements ("I prefer…"), or explicit triggers ("remember that…")
2. **Extract the learning** — identify what was wrong and what the correct behavior is, structured as a key-value pair:
   - Category: `tone`, `formatting`, `factual`, `preference`, `workflow`
   - Before: what Zoe did wrong
   - After: what the user wants instead
3. **Check for duplicates** — use `GET /api/learnings` before proposing a new learning
4. **Acknowledge to the user** — respond with a brief acknowledgment (e.g., "Got it, I'll remember that for next time"); the self-learning pipeline handles actual persistence
5. **Apply in future** — before generating responses, retrieve relevant learnings from `GET /api/learnings` and adjust behavior accordingly

## Example

**User:** "Actually, I prefer bullet points over paragraphs when you summarize things."

**Steps:**
- Detect preference signal ("I prefer")
- Extract: category=`formatting`, before=`paragraph summaries`, after=`bullet point summaries`
- Let the self-learning pipeline persist the learning with user ownership; this skill only uses the listed review endpoints
- Respond: "Noted — I'll use bullet points for summaries from now on."

## Security

- Only trusted sources can create learnings (Trust Gate integration)
- Owner corrections auto-confirm via `POST /api/learnings/{id}/confirm`
- Trusted contact corrections require user review before confirming
- Learnings expire after 30 days without confirmation — dismiss stale ones via `POST /api/learnings/{id}/dismiss`