Self Improvement
Turn agent mistakes, user corrections, and integration failures into durable `.learnings/` logs so solo builders compound reliability across sessions.
Overview
Self-improvement is a journey-wide agent skill that captures errors, corrections, and capability gaps into `.learnings/` markdown—usable whenever a solo builder needs to record what went wrong before committing to the ne
Install
npx skills add https://github.com/pskoett/self-improving-agent --skill self-improvementWhat is this skill?
- Auto-initializes `.learnings/` with LEARNINGS.md, ERRORS.md, and FEATURE_REQUESTS.md without overwriting existing files
- Structured categories: correction, insight, knowledge_gap, and best_practice
- Triggers on command failures, user pushback, missing capabilities, and outdated agent knowledge
- Encourages reviewing learnings before major tasks
- Feeds downstream promotion into project memory for lasting fixes
- 3 default markdown logs: LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md
- 4 learning categories: correction, insight, knowledge_gap, best_practice
Adoption & trust: 8.3k installs on skills.sh; 11 GitHub stars; 3/3 security scanners passed (skills.sh audits).
What problem does it solve?
You lose the same debugging lessons and user corrections every time the agent session resets, so mistakes repeat across build, ship, and operate work.
Who is it for?
Solo builders running long agent sessions on real codebases who want a zero-infra habit for post-mortems and correction tracking.
Skip if: Teams that already mandate a centralized incident or knowledge base with strict access controls and no repo-local markdown.
When should I use this skill?
Command or operation fails unexpectedly; user corrects the agent; user requests missing capability; external API or tool fails; knowledge is outdated; better approach found; review learnings before major tasks.
What do I get? / Deliverables
After runs, you get categorized learnings and error logs on disk that you can review before big tasks and promote into durable project memory.
- Initialized `.learnings/` tree
- Structured entries in LEARNINGS.md, ERRORS.md, or FEATURE_REQUESTS.md
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
Log a third-party API timeout and the workaround before retrying the integration.
Record a flaky test failure and the user’s preferred assertion fix before the next CI run.
Capture a wrong deploy command correction before shipping an update.
Append production-like errors from local repro steps into ERRORS.md for later triage.
File a feature request when the user asks for a capability the skill set does not yet cover.
How it compares
Use as procedural memory for the agent instead of hoping chat history alone will prevent repeat failures.
Common Questions / FAQ
Who is self-improvement for?
Indie and solo developers using coding agents who want repo-local logs of what failed, what the user corrected, and what capabilities were requested—without a separate ticketing tool.
When should I use self-improvement?
Use it when a command or integration fails, the user corrects the agent, a requested capability is missing, knowledge is stale, or you discover a better approach; also review `.learnings/` before major build, ship, launch, or operate milestones.
Is self-improvement safe to install?
It mainly creates and appends markdown under `.learnings/`; review the Security Audits panel on this page and avoid logging secrets or credentials into those files.
SKILL.md
READMESKILL.md - Self Improvement
# Self-Improvement Skill Log learnings and errors to markdown files for continuous improvement. Coding agents can later process these into fixes, and important learnings get promoted to project memory. ## First-Use Initialisation Before logging anything, ensure the `.learnings/` directory and files exist in the project or workspace root. If any are missing, create them: ```bash mkdir -p .learnings [ -f .learnings/LEARNINGS.md ] || printf "# Learnings\n\nCorrections, insights, and knowledge gaps captured during development.\n\n**Categories**: correction | insight | knowledge_gap | best_practice\n\n---\n" > .learnings/LEARNINGS.md [ -f .learnings/ERRORS.md ] || printf "# Errors\n\nCommand failures and integration errors.\n\n---\n" > .learnings/ERRORS.md [ -f .learnings/FEATURE_REQUESTS.md ] || printf "# Feature Requests\n\nCapabilities requested by the user.\n\n---\n" > .learnings/FEATURE_REQUESTS.md ``` Never overwrite existing files. This is a no-op if `.learnings/` is already initialised. Do not log secrets, tokens, private keys, environment variables, or full source/config files unless the user explicitly asks for that level of detail. Prefer short summaries or redacted excerpts over raw command output or full transcripts. If you want automatic reminders or setup assistance, use the opt-in hook workflow described in [Hook Integration](#hook-integration). ## Quick Reference | Situation | Action | |-----------|--------| | Command/operation fails | Log to `.learnings/ERRORS.md` | | User corrects you | Log to `.learnings/LEARNINGS.md` with category `correction` | | User wants missing feature | Log to `.learnings/FEATURE_REQUESTS.md` | | API/external tool fails | Log to `.learnings/ERRORS.md` with integration details | | Knowledge was outdated | Log to `.learnings/LEARNINGS.md` with category `knowledge_gap` | | Found better approach | Log to `.learnings/LEARNINGS.md` with category `best_practice` | | Simplify/Harden recurring patterns | Log/update `.learnings/LEARNINGS.md` with `Source: simplify-and-harden` and a stable `Pattern-Key` | | Similar to existing entry | Link with `**See Also**`, consider priority bump | | Broadly applicable learning | Promote to `CLAUDE.md`, `AGENTS.md`, and/or `.github/copilot-instructions.md` | | Workflow improvements | Promote to `AGENTS.md` (OpenClaw workspace) | | Tool gotchas | Promote to `TOOLS.md` (OpenClaw workspace) | | Behavioral patterns | Promote to `SOUL.md` (OpenClaw workspace) | ## OpenClaw Setup (Recommended) OpenClaw is the primary platform for this skill. It uses workspace-based prompt injection with automatic skill loading. ### Installation **Via ClawdHub (recommended):** ```bash clawdhub install self-improving-agent ``` **Manual:** ```bash git clone https://github.com/peterskoett/self-improving-agent.git ~/.openclaw/skills/self-improving-agent ``` Remade for openclaw from original repo : https://github.com/pskoett/pskoett-ai-skills - https://github.com/pskoett/pskoett-ai-skills/tree/main/skills/self-improvement ### Workspace Structure OpenClaw injects these files into every session: ``` ~/.openclaw/workspace/ ├── AGENTS.md # Multi-agent workflows, delegation patterns ├── SOUL.md # Behavioral guidelines, personality, principles ├── TOOLS.md # Tool capabilities, integration gotchas ├── MEMORY.md # Long-term memory (main session only) ├── memory/ # Daily memory files │ └── YYYY-MM-DD.md └── .learnings/ # This skill