leprachuan/self-improving-agent
self-improving-agent/SKILL.md
Capture corrections, errors, feature requests, and recurring best practices in a local `.learnings/` workspace so future agent runs can avoid repeating mistakes. Use when a command fails, a user corrects the agent, a missing capability is requested, a better repeatable approach is discovered, or before major work to review prior lessons.
data-ai
Updated Apr 21, 2026
$ install --global
skillsauthnpx skillsauth add leprachuan/pot-o-skills self-improving-agentInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 30, 2026, 1:09 PM7.6s1 file scanned
SKILL.md
- name:
- self-improving-agent
- description:
- Capture corrections, errors, feature requests, and recurring best practices in a local `.learnings/` workspace so future agent runs can avoid repeating mistakes. Use when a command fails, a user corrects the agent, a missing capability is requested, a better repeatable approach is discovered, or before major work to review prior lessons.
Self-Improving Agent
Use this skill to keep a lightweight local learning loop for Wee Orchestrator work and related repos.
Quick Start
Initialize a learning workspace in the current project root:
./scripts/init_learnings.sh
That creates:
.learnings/
├── LEARNINGS.md
├── ERRORS.md
└── FEATURE_REQUESTS.md
Do not overwrite existing files. Keep entries short, specific, and sanitized.
Wee Orchestrator Adaptation
This skill is adapted for Wee Orchestrator rather than OpenClaw.
- Use
.learnings/in the repo or agent workspace you are actively changing. - For Wee Orchestrator dev work, prefer
/opt/n8n-copilot-shim-dev/.learnings/on the dev host. - For local agent repos like
/opt/wee-dev,/opt/wee-qa, or/opt/skills/<skill-name>, keep.learnings/inside that repo. - Do not store secrets, tokens, raw environment files, or full logs. Store summaries or redacted excerpts.
When to Log
| Situation | Destination |
|---|---|
| Unexpected command or tool failure | .learnings/ERRORS.md |
| User correction or outdated assumption | .learnings/LEARNINGS.md |
| Missing capability request | .learnings/FEATURE_REQUESTS.md |
| Better recurring implementation pattern | .learnings/LEARNINGS.md |
| Repeated bug pattern from multiple tasks | Update existing learning and increase recurrence metadata |
Before Major Work
Before a substantial task, scan local learnings:
grep -n "^## \[" .learnings/*.md 2>/dev/null
If you are working in a familiar area, also search by topic:
grep -Rni "auth\|scheduler\|queue\|frontend" .learnings/ 2>/dev/null
Promotion Targets
When a learning is stable and broadly useful, promote it out of .learnings/:
- Project conventions →
CLAUDE.md - Agent workflow guidance →
AGENTS.md - Copilot-specific reminders →
.github/copilot-instructions.md - Cross-session durable lessons → Total Recall
agent_lessons - Repository-wide implementation facts → repo memory/store-memory workflow
For Wee Orchestrator specifically, the most common promotion targets are /opt/AGENTS.md, /opt/CLAUDE.md, and the relevant repo-local AGENTS.md.
Entry Guidelines
Use IDs like LRN-YYYYMMDD-001, ERR-YYYYMMDD-001, and FEAT-YYYYMMDD-001.
Always include:
- timestamp
- priority
- status
- area
- concise summary
- concrete suggested action
- related files when known
Prefer updating an existing entry over duplicating it. Link related items with See Also.
Status Values
Use these statuses consistently:
pendingin_progressresolvedwont_fixpromotedpromoted_to_skill
Hook / Reminder Support
This skill includes small helper scripts:
scripts/activator.sh— reminder at prompt startscripts/error-detector.sh— reminder after shell/tool failures
For Claude-style environments with hooks, see references/wee-orchestrator-integration.md.
For Copilot CLI, add a short reminder to AGENTS.md or copilot-instructions.md instead of relying on hooks.
Resources
references/examples.md— example entries for each log typereferences/wee-orchestrator-integration.md— setup guidance for local agent reposreferences/upstream.md— upstream source references and adaptation notesassets/— starter templates for.learnings/files
Related Skills
leprachuan/github-kanban-board
data-ai
VerifiedTrustedCommunity
Interactive GitHub issues kanban board with agent assignments, due dates, and glassmorphism theming. Fully configurable for any GitHub repository.
SKILL.mdUpdated Jun 7, 2026
leprachuan/wee-canvas-todos
data-ai
VerifiedTrustedCommunity
Interactive TODO board for Wee Canvas. Displays TODOs from both GitHub Issues (leprachuan/fosterbot-home) and flat files in two views: list and kanban. Features filtering, drag-and-drop status changes, quick-add, and auto-refresh every 30 seconds. Use when Foster asks to "show TODOs", "open TODO board", "view my tasks", or "TODO kanban".
SKILL.mdUpdated Apr 21, 2026
leprachuan/webssh-terminal
tools
VerifiedTrustedCommunity
Web-based terminal tools for Wee Canvas: remote SSH terminal (WebSSH) and local bash terminal (ttyd). Embeds interactive terminal panels in Wee Canvas iframes. Use when the user asks for a 'web terminal', 'local terminal', 'browser SSH', 'webssh', or wants to interact with a host through the WebUI canvas. For browser windows, see the browser-window skill.
SKILL.mdUpdated Apr 21, 2026
leprachuan/webex-notify
development
VerifiedTrustedCommunity
Use when you need to send WebEx notifications to flipkey-home-bot - supports markdown formatting, auto-retry with backoff, rate limiting, and message history tracking
SKILL.mdUpdated Apr 21, 2026