# Deliberate Practice

> Deliberate Practice

- Skill: `sethdford/deliberate-practice` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sethdford/deliberate-practice`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sethdford/deliberate-practice/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sethdford (https://skillmd.com/u/sethdford)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sethdford/deliberate-practice

---

# Deliberate Practice

Design training like **Anders Ericsson’s** research: isolate sub-skills, work at the edge of current ability, get **immediate** feedback, repeat with **variation**. Volume without focus mostly rehearses existing habits.

## When to Use
- The user wants to improve a skill (code, writing, speaking, debugging, design) and is stuck “practicing” without measurable movement.

## Workflow
1. **Decompose** the skill into specific sub-skills (e.g., “read stack traces fast,” “name variables clearly,” “give feedback in one pass”).
2. **Pick one weak sub-skill** per session—not comfortable strengths.
3. **Set a performance goal**, not a time goal (e.g., “three correct diagnoses from stripped logs,” not “30 minutes of practice”).
4. **Design a tight loop**: attempt → criteria → feedback in seconds or minutes (tests, rubric, peer, recording review).
5. **Vary constraints**: harder inputs, shorter timebox, different domain—same sub-skill.

Normalize **productive discomfort** as signal you are past autopilot. Help the user iterate the plan when feedback is noisy or delayed.

## Examples
**Example 1:** Debugging → 10-minute drills on one bug class with “predict before open file” then diff against ground truth.

**Example 2:** Writing → rewrite the same paragraph under opposing constraints (half length, then twice empathy) with a checklist score each round.