kkkkhazix/khazix-skills
4 skills6.2k installs56.3k starsGitHub
Install
npx skills add https://github.com/kkkkhazix/khazix-skillsSkills in this repo
1Khazix WriterKhazix-writer is an agent skill that writes WeChat 公众号 long-form articles in the personal voice of 数字生命卡兹克—described as “有见识的普通人在认真聊一件打动他的事.” Solo builders and creators install it when they have素材 (product briefs, news, PDFs, voice transcripts, or bullet outlines) and need a full draft or continuation that sounds like a curious human sharing real experience, not generic AI prose. The workflow starts with understanding素材 and running HKR topic QA (interesting hook, teachable insight, emotional resonance); weak angles get flagged for realignment with the user. It encodes Khazix’s editorial values—curiosity-first, 活人感, honesty about flaws and limits, and refusing traffic that betrays conviction—while keeping AI in a support role: evidence, analogies, structure, and polish rather than replacing the author’s judgment. It fits indie operators building an AI-curious audience in Chinese, not teams needing corporate comms or English-only blogs.2kinstalls2Hv Analysishv-analysis is a Horizontal-Vertical Analysis Framework packaged as structured JSON your agent fills in while researching a product, company, concept, or person. Solo and indie builders install it when they need more than ad-hoc notes: the vertical axis narrates origin through key milestones and phased evolution, while the horizontal axis forces side-by-side competitive and positioning comparison at the current moment. The schema prompts for motives, abandoned paths, external shocks, and why each node mattered—material you can reuse in positioning docs, investor one-pagers, or validate-phase scope decisions. It does not fetch live market data; it disciplines how you organize what you find. Use it early in the journey when choosing what to build or how to differentiate, and again when refreshing competitive context before launch or growth bets.1.8kinstalls3Neat Freakneat-freak is a compact agent-memory and configuration path cheat sheet for solo and indie builders who juggle Claude Code, OpenAI Codex, OpenClaw, and OpenCode. Instead of guessing where MEMORY.md, CLAUDE.md, AGENTS.md, or platform skill folders live, you run a structured first-pass inventory using the tables in the skill. It clarifies that Claude Code uses encoded project paths and indexed memory files, while Codex and OpenClaw lean on layered AGENTS.md-style instructions without a separate memory index. OpenClaw’s skill resolution order (workspace over bundled) and OpenCode’s project-scoped .opencode layout are spelled out so you do not overwrite the wrong layer. Use it when onboarding a repo to an agent, migrating between tools, or untangling duplicate skills after copying configs. It does not move files for you—it orients you so later hygiene, sync, and skill installs hit the correct directory on the first try.1.4kinstalls4Aihotaihot is an agent skill that turns AI HOT’s public JSON APIs into readable Chinese markdown digests for solo builders who need same-day AI industry signal instead of hallucinated or outdated training knowledge. It targets anyone shipping with Claude Code, Cursor, Codex, or compatible agents when they ask about AI 日报, hotspots, OpenAI/Anthropic/Google releases, or casual “what happened in AI” questions in an LLM context. The workflow is curl-first: set a Chrome-like User-Agent for all /api/public/* calls, then fetch daily, featured, or time-ranged endpoints and synthesize titles, links, and context. Because the site blocks bare curl user agents, the skill documents that pitfall up front so agents do not misreport the service as down. It is lightweight integration work—network and shell only—with no secrets. Use it during early research, validation conversations, and ongoing operate/grow check-ins when market moves affect your roadmap.992installs