knopki's curated collection of agent skills.
This repository also configures its own agents. See what agent-skills tells them →
Skill Claude CodeCodex
Use when the user asks to "write a prompt for DeepSeek", "compose a DeepSeek prompt", wants to send a task to a DeepSeek model, needs to launch a DeepSeek subagent. Also use when explicitly named: DeepSeek. Trigger on any request to draft, refine, or assemble a prompt targeting a DeepSeek-family model — for either…
Skill Claude CodeCodex
Use when a task needs live web research: web search for URLs, fetching a page as clean markdown, or crawling a site into markdown. Trigger on "search the web", "fetch/read this URL", "crawl this site/docs", current info beyond training data, bot-walled/JS pages, PDF text extraction.
Skill Claude CodeCodex
Use when the user asks to "write a prompt for GLM", "compose a GLM prompt", wants to send a task to a GLM model, needs to launch a GLM subagent. Also use when explicitly named: GLM. Trigger on any request to draft, refine, or assemble a prompt targeting a GLM-family model — for either interactive use or subagent…
Skill Codex
Use when the user asks about GRACE semantic code markup, asks to set up project, review. Triggers: GRACE init, GRACE review, or when code already contains #region MODULECONTRACT markers.
Skill Claude CodeCodex
Use when a task needs to find real code patterns in public GitHub repositories via the grep.app MCP CLI — for example, verifying how a library, framework, config file, or language feature is actually used in production. Trigger on requests to search code (not documentation), find concrete usage examples, or inspect…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: