self-improvement

Record runtime learnings, errors, and feature requests into structured markdown logs.

Updated Mar 23, 2026
One-click install
npx skills add https://github.com/giruclawbot/giru-backup --skill self-improvement-giruclawbot
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-improvement
Source: https://github.com/giruclawbot/giru-backup/tree/main/skills/self-improving-agent
Command: npx skills add https://github.com/giruclawbot/giru-backup --skill self-improvement-giruclawbot

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Sessions and developer workflows frequently lose corrections, failed-command context, and discovered best practices because there's no consistent, machine-readable place to record them. This Skill provides a structured way to capture transient insights, error details, and feature requests so they can be reviewed, promoted to project memory, and acted on by agents or humans.

Core Features & Use Cases

  • Structured logging: Templates and ID conventions for LEARNINGS, ERRORS, and FEATURE_REQUESTS to ensure consistency.
  • Agent integration: Guidance and optional hooks for OpenClaw, Claude, Codex, and Copilot workflows to surface reminders and detect failures.
  • Promotion workflow: Rules for when to promote learnings into CLAUDE.md, AGENTS.md, SOUL.md, or other project memory targets.
  • Use cases: Log a failing CI command, capture a user correction during a session, record a requested capability for future work, or track recurring patterns for automation.

Quick Start

Ask the agent to log the recent failed command and its suggested fix to .learnings/ERRORS.md with a timestamp, priority, and reproduction steps.

Frequently Asked Questions about self-improvement

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I log runtime errors and user corrections into structured markdown files?

Capture runtime errors and user corrections by applying templated LEARNINGS, ERRORS, and FEATURE_REQUESTS formats into structured markdown files with timestamps, metadata, unique IDs, and resolution fields for automated processing.

What is the best way to capture failed command context during developer workflows?

The best way to capture failed command context is recording error details, reproduction steps, and suggested fixes into structured markdown log files to prevent losing transient insights across developer sessions.

Can I use structured learning logs with OpenClaw and Claude workflows?

Yes, structured learning logs integrate with OpenClaw, Claude, Codex, and Copilot workflows through guidance and optional hooks that surface reminders and detect tool or command failures during agent automation.

When should I promote logged learnings into project memory files like CLAUDE.md?

Promote logged learnings into project memory files like CLAUDE.md, AGENTS.md, or SOUL.md when applying promotion rules to move reviewed entries from transient logs into permanent project memory targets for agents.

How do I record feature requests for future agent automation work?

Record feature requests for future agent automation by using the FEATURE_REQUESTS template format with unique IDs, timestamps, and metadata to ensure requested capabilities are tracked and acted on by agents or humans.