hwamony_skills

A reusable skill library for turning vague AI work into structured, inspectable workflows.

hwamony_skills collects reusable hwamony- skills across agent building, prompt engineering, research workflows, ideation, naming, skill improvement, and careful human-facing conversations. The collection is built for people who want more than one-shot prompts: clearer workflows, better artifacts, stronger handoffs, and a repo you can browse skill by skill without guessing what each folder is for.

If you only try a few first, start with the flagship skills below.

🚀 Start Here

Copy one of these prompts into your agent:

• Use $hwamony-prompt-architect to turn this messy system prompt into a production-ready prompt spec with schema, tool rules, and model guidance.
• Use $hwamony-desk-research to turn this broad market question into a scoped research folder with saved sources and a memo.
• Use $hwamony-creative-thinking to map the real problem behind this brief and keep one surprising but viable direction alive in the shortlist.
• Use $hwamony-skill-lab to test one bounded mutation of this skill and keep it only if evaluator-majority says it beats baseline.
• Use $hwamony-skill-showcase to rewrite this internal skill so GitHub visitors instantly understand it and know what to try first.

🔎 What Lives Here

• AI skills for agent building, prompt engineering, and system design
• research workflow skills for desk research, competitor scans, and source-backed memos
• ideation skills for problem framing, idea generation, naming, positioning, and launch copy
• skill improvement and skill showcase workflows for hardening and publishing reusable skills
• structured human-facing conversation skills with visible safety boundaries

🧭 Category Map

• agent_builder: design, implement, evaluate, extend, and publish agent systems or prompt-driven workflows
• research: turn broad questions into scoped research projects with saved evidence and reusable artifacts
• ideation: move from fuzzy briefs to concept maps, stronger options, names, and launch angles
• mind: structured human-facing conversations for self-reflection and behavior analysis with visible safety boundaries
• skill_lab: evaluate and improve skills through bounded experiments, fixed cases, and promotion-safe loops

⭐ Why Star This Repo

• the skills are built as reusable workflows, not random prompt snippets
• many of them produce files, folders, or handoff artifacts instead of chat-only output
• the collection covers the full loop from ideation to research to agent building to publishable docs
• each skill is documented so you can judge fit quickly instead of reading a giant monorepo blind

🧩 What You Can Do With It

• turn a fragile prompt into a schema-first prompt system
• turn a broad question into staged desk research with saved sources
• turn a vague problem into a shortlist of stronger solution directions
• turn a service brief into market-aware names and launch copy
• turn a promising but messy skill into fixed cases, dated runs, backups, and evidence-backed improvements
• turn a good internal skill into a public-facing README, stronger GitHub metadata, and a launch-ready showcase

✨ Why This Repo Feels Different

• these are not one-shot prompt snippets; most skills define a reusable workflow
• many skills save artifacts to disk instead of leaving the work trapped in chat
• the collection spans ideation, research, agent building, and careful human-facing conversations
• several skills include scripts, references, or implementation scaffolds alongside the skill text
• the repo is organized so you can inspect one skill at a time without learning the whole system first

⭐ Flagship Skills

• hwamony-prompt-architect Design prompts as reusable systems with routing, schemas, tools, and production constraints.
• hwamony-desk-research Turn one-shot searching into staged desk research with saved sources and reusable artifacts.
• hwamony-creative-thinking Turn a fuzzy brief into core concepts, a visual map, distant-combination ideas, and a shortlist that still keeps one surprising lane alive.
• hwamony-skill-lab Turn a promising skill into dated experiments, fixed cases, supervisor-planned mutations, evaluator-agent judgments, and safe promotion.
• hwamony-skill-showcase Turn an internal skill into a public-facing README, stronger discoverability signals, and launch-ready GitHub copy.

🧭 Pick A Starting Point

• If your prompt keeps breaking when the task gets real: start with hwamony-prompt-architect
• If your research questions are broad and you need saved evidence: start with hwamony-desk-research
• If the team is brainstorming the wrong thing: start with hwamony-creative-thinking
• If your skill is promising but you still cannot prove it is improving: start with hwamony-skill-lab
• If you need a name and launch copy that still fits the live market: start with hwamony-service-naming-copy
• If your skill works but still looks private or hard to discover on GitHub: start with hwamony-skill-showcase
• If you are building an agent from scratch: start with hwamony-agent-system-builder

🗂️ Full Collection

mind

• hwamony-abc-fa
• hwamony-cbt-chats

ideation

• hwamony-creative-thinking
• hwamony-service-naming-copy

research

• hwamony-desk-research
• hwamony-debate-panel

agent_builder

• hwamony-agent-system-builder
• hwamony-agent-system-evaluator
• hwamony-agent-system-implementer
• hwamony-prompt-architect
• hwamony-search-skills
• hwamony-skill-showcase

skill_lab

• hwamony-skill-lab

This category now owns skill-improvement loops that used to be described more loosely as harness workflows. If the job is "make this skill reliably better over time," start here rather than in agent_builder.

📝 Repo Notes

• hwamony-convergent-thinking and hwamony-divergent-thinking are intentionally excluded.
• skills are shared here without local .env, .DS_Store, or autoresearch-* artifacts
• original code and prose use different licenses; see below

⚖️ Copyright and Licensing

Copyright (c) 2026 Hwamony.

This repository uses a split license model:

• original code inside scripts/ directories is licensed under the MIT License
• original SKILL.md, README.md, and other prose documentation are licensed under CC BY 4.0

See:

• LICENSE
• LICENSES/MIT.txt
• LICENSES/CC-BY-4.0.md

📎 Notes

• each skill is meant to be inspectable on its own via its local README.md
• category READMEs give a faster entry point if you want to browse by workflow instead of by file tree

🙏 Attribution

If you reuse the skill text or documentation, please provide clear attribution to Hwamony and note any changes you made.