Agent Skills › heyron-ai/agent-library › self-improvement

self-improvement

GitHub

用于记录错误、用户纠正和新发现的自动化技能，通过结构化日志实现持续学习与优化。

skills/self-improvement/SKILL.md heyron-ai/agent-library

Trigger Scenarios

命令执行失败用户进行纠正或提供新信息

Install

npx skills add heyron-ai/agent-library --skill self-improvement -g -y

More Options

Use without installing

npx skills use heyron-ai/agent-library@self-improvement

指定 Agent (Claude Code)

npx skills add heyron-ai/agent-library --skill self-improvement -a claude-code -g -y

安装 repo 全部 skill

npx skills add heyron-ai/agent-library --all -g -y

预览 repo 内 skill

npx skills add heyron-ai/agent-library --list

SKILL.md

Frontmatter

{
    "name": "self-improvement",
    "description": "Log learnings, errors, and feature requests for continuous improvement. Use when: (1) a command fails, (2) user corrects you, (3) user requests something you can't do, (4) you discover outdated knowledge, (5) you find a better approach to a recurring task."
}

Self-Improvement Skill

Log mistakes, corrections, and discoveries to .learnings/ files so you get smarter over time.

Setup

mkdir -p ~/workspace/.learnings

Quick Reference

Situation	Action
Command fails	Log to `.learnings/ERRORS.md`
User corrects you	Log to `.learnings/LEARNINGS.md`
User wants missing feature	Log to `.learnings/FEATURE_REQUESTS.md`
Found better approach	Log to `.learnings/LEARNINGS.md`
Knowledge was outdated	Log to `.learnings/LEARNINGS.md`

When to Promote

When a learning proves broadly useful, promote it to a permanent file:

Learning Type	Promote To
Behavioral patterns	`SOUL.md`
Workflow improvements	`AGENTS.md`
Tool gotchas	`TOOLS.md`

Logging Format

Learning Entry

Append to .learnings/LEARNINGS.md:

## [LRN-YYYYMMDD-001] category

**Logged**: 2026-03-05
**Priority**: low | medium | high | critical
**Status**: pending

### Summary
One-line description

### Details
What happened, what was wrong, what's correct

### Suggested Action
Specific fix or improvement

Error Entry

Append to .learnings/ERRORS.md:

## [ERR-YYYYMMDD-001] what_failed

**Logged**: 2026-03-05
**Priority**: high
**Status**: pending

### Summary
What failed

### Error

Actual error message


### Context
What you were trying to do

### Suggested Fix
How to prevent this next time

Feature Request Entry

Append to .learnings/FEATURE_REQUESTS.md:

## [FEAT-YYYYMMDD-001] capability_name

**Logged**: 2026-03-05
**Priority**: medium
**Status**: pending

### Requested Capability
What the user wanted

### Suggested Implementation
How this could work

Detection Triggers

Log automatically when you notice:

Corrections: "No, that's wrong...", "Actually...", "You're wrong about..."
Feature requests: "Can you also...", "I wish you could...", "Is there a way to..."
Errors: Non-zero exit codes, exceptions, unexpected output
Knowledge gaps: User provides info you didn't know

Resolving Entries

When fixed, update the entry:

Change **Status**: pending → **Status**: resolved
Add what was done to fix it

Review

Check your learnings periodically:

# Count pending items
grep -c "Status.*pending" ~/workspace/.learnings/*.md

# Find high priority items
grep -B3 "Priority.*high" ~/workspace/.learnings/*.md

Best Practices

Log immediately — context is freshest right after the issue
Be specific — future you needs to understand quickly
Suggest concrete fixes — not just "investigate"
Promote aggressively — if it's broadly useful, put it in SOUL.md/AGENTS.md/TOOLS.md