Skip to content

Skillselionwith God's help

By phase

IdeaResearch & spot demand
ValidateScope & prototype
BuildFrontend, backend, MCP
ShipTest, review, deploy
LaunchDistribution, SEO, ASO
GrowAnalytics & retention
OperateMonitoring & uptime

By type

Skills
Marketplaces
MCP servers
Workflows

Skills Marketplaces MCP Learn Brief

Browse all tools →By phase

Idea Validate Build Ship Launch Grow Operate

Explore

Skills
Marketplaces
MCP
Learn
Brief

This week for builders

Five minutes, every Monday — the tools, releases and tactics for shipping solo.

Email address

unsubscribe anytime.

Skillselion

An independent, curated directory of AI-coding tools for solo builders - skills, MCP servers, marketplaces and workflows for Claude Code, Codex and Cursor, organized by the indie build journey from idea to operate.

ResourcesBrowse Skills Browse MCP Servers Browse Marketplaces Browse Workflows Best-of shortlists Glossary State of AI Skills 2026 Weekly Brief

CommunityAbout Learn Feedback Privacy Policy Terms Cookie Policy Accessibility Cookie settings

Last updated June 10, 2026 · © 2026 Skillselion. Skillselion is an independent project and is not affiliated with Anthropic, OpenAI, Cursor, Claude, Claude Code, or Codex.

Skills/PSKOETT/SELF IMPROVING AGENT

pskoett/self-improving-agent

1 skill · 8.3k installs · 11 stars · GitHub

Install

npx skills add https://github.com/pskoett/self-improving-agent

Skills in this repo

1Self ImprovementSelf-improvement is an agent skill that turns messy session feedback into a lightweight learning system for solo builders shipping with Claude Code, Cursor, or similar tools. When a shell command fails, an API integration breaks, the user corrects the agent, or a recurring workflow gets a better approach, the skill ensures `.learnings/` exists and appends structured notes so nothing disappears after the next chat clear. It separates corrections, insights, knowledge gaps, and best practices from raw error traces and feature asks, which makes later triage and automation predictable. Because triggers span build, ship, and operate work, it behaves as journey-wide process infrastructure rather than a one-off linter. Reviewing learnings before large refactors or launches reduces repeated mistakes without requiring a separate observability stack—ideal for indies who want compounding agent quality from plain markdown in the repo root.8.3kinstalls