published by cskwork

25 listings on this page, in order of arrival. Each one has its own page with README, repository facts and source links.

1. supergoal ★ 4
   Objective-driven dev router. Use for "fix bug", "add feature", "prototype", "QA/verify", or "code review".
2. handoff ★ 1
   Handoff — save state to pause, resume, or transfer work. Use when: "handoff", "pause here", "continue later".
3. reddit-poster ★ 1
   reddit-poster — human-style Reddit posts. Use when: "post on Reddit", "share on r/X", "/reddit-poster".
4. browser-qa ★ 1
   Browser QA with reviewable YAML scenarios. Use when the user says QA, gives a URL to verify, or records a test.
5. superadvisor ★ 1
   Router to the best super* skill, persona, or gstack workflow. Use when a task spans domains or no skill owns it.
6. postgres-intelligence ★ 2
   PostgreSQL for LLM agents: inspect schemas, run safe SQL, translate natural language to queries. Use when connecting to a database.
7. superoffice ★ 1
   Business-document maker (docx/pptx/xlsx/hwpx/pdf) with company branding and gate-checked output. Use for /superoffice.
8. fizzy ★ 1
   Operate a self-hosted Fizzy kanban board via fizzy-cli. Use when creating cards or migrating Jira issues to Fizzy.
9. superpm ★ 1
   PM artifacts verified by a critic. Use for PRDs, strategy, OKRs, roadmaps, specs, research, analytics, or GTM.
10. streamdeck-agent-cockpit ★ 1
    Turn a Stream Deck into a standalone local cockpit for Claude Code, Codex, Pi, JCode, and other terminal agents: launch, focus, resume, and
11. claude-code-workflow-cheatsheet ★ 1
    Generate companyAnnouncements cheat sheets for installed command harnesses. Use when setting up workflow guidance.
12. myfocus
    myfocus — top next task across agent sessions. Use when: "what next", "next task", "다음 작업", "prioritize".
13. skill-curator
    Curate local Agent Skills libraries: inventory, validate, deduplicate, pin, archive, and restore SKILL.md packages across coding harnesses.
14. superdesign
    Web/mobile UI design past an anti-slop gate. Use for /superdesign, "design X", "make this beautiful", or "redesign UI".
15. superloop
    Use /loop to verify a delivery or advance a project one ticket per tick against acceptance criteria.
16. web-legacy-compass
    Trace real legacy-web user flows before changing code. Use when request order and DB reads/writes must be correlated.
17. debug-code
    Diagnose and fix hard bugs using runtime evidence and tracing. Use when debugging broken or intermittent behavior.
18. codex-cli
    Delegate review, implementation, or image generation to the Codex CLI for a second opinion. Use when asking codex.
19. superhacker
    Authorized-security router. Use for /superhacker, pentest, vulnerability triage, red team, hardening, or CTF.
20. learn-codebase
    learn-codebase — intent-anchored code reading. Use when learning unfamiliar code, reviewing AI PRs, or spotting drift.
21. clean-code
    Refactor legacy code for readability without changing behavior. Use when cleaning up or reducing technical debt.
22. skill-ab-eval
    skill-ab-eval — A/B test skill lift, compare CLI agents. Use when: "evaluate skill", "skill A/B", "compare CLIs".
23. context-diet
    Cut Claude Code system-prompt bloat by disabling unused tools/skills in settings.json. Use when reducing token usage.
24. stitch-landing
    stitch-landing — README to GitHub Pages landing. Use when: "landing page", "github pages", "랜딩 페이지".
25. ssh-llm-connect
    ssh-llm-connect — safe SSH for agents to read remote logs. Use when: "ssh to prod", "read logs", "서버 봐야 함".