菜单
Import and distill leaked GPT prompts from linexjlin/GPTs into agent knowledge files. Fetches directly from GitHub, distills domain knowledge via agent-refine-knowledge, logs source to claude-secrets catalog.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
YouTube transkript, ozet, arastirma ve soru-cevap araci. transcriptr'in yerel headless API'sini (localhost:8080) kullanir; her projeden cagrilabilir. Triggers: transcriptr, youtube transkript, video transkript, videoyu transkript et, video ozet, videoyu ozetle, youtube ozet, konu arastir, youtube arastir, video arastir, videolardan ozet, video soru cevap, transcript, summarize video, research youtube.
Tam otonom gelistirme dongusu — analysis, sprint, parallel tasks, PR/review loop, merge, summary. Focus mode ile belirli alanlara odaklanir. Kendi repeat'i dahil. Triggers: forge, forge all, tam dongü, full cycle.
Bekleyen asset-browser upload'larini incele — her biri icin onay/red karari ver, onaylananlari post-process edip runtime asset dizinine tasi, commit+push. Triggers: asset review, asset incele, bekleyenleri incele, asset review workflow, waiting review.
Asset browser lokal dev sunucusu başlat ve tarayıcıda aç. Triggers: asset browser local, asset browser dev, local asset browser, lokal browser
Asset browser Vercel canlı URL'ini tarayıcıda aç. Triggers: asset browser online, asset browser vercel, online asset browser, canlı browser
Asset browser paket kodunu projeye çek + redeploy. Triggers: asset browser update, browser güncelle, update asset browser, asset browser paket güncelle
| name | gpt-import |
| description | Import and distill leaked GPT prompts from linexjlin/GPTs into agent knowledge files. Fetches directly from GitHub, distills domain knowledge via agent-refine-knowledge, logs source to claude-secrets catalog. |
| invocation | /gpt-import |
| argument-hint | <agent-id> [--gpt=<gpt-name>] [--search=<query>] |
/gpt-import /gpt-import --gpt=logo-creator /gpt-import --search=seo
Fetches leaked custom GPT instructions from linexjlin/GPTs GitHub repo and distills
the domain knowledge into the agent's knowledge/ directory.
Resolve agent — Find agent by ID or name in ~/Projects/claude-config/config/agent-registry.json.
Get agent's category and capabilities.
Browse available GPTs — Fetch file list from GitHub API:
gh api repos/linexjlin/GPTs/contents/prompts --jq '.[].name' 2>/dev/null
If --gpt=<name> specified: fetch that specific file directly.
If --search=<query>: filter list by query, show top 10 matches to user.
Otherwise: show list filtered by agent's category keywords, ask user to pick.
Fetch selected GPT content:
gh api repos/linexjlin/GPTs/contents/prompts/<filename> --jq '.content' | base64 -d
Distill — Call /agent-refine-knowledge skill with:
Write to knowledge/ — The refine skill writes to agent's knowledge/ dir.
Log to sources catalog — Append entry to ~/Projects/claude-secrets/sources-catalog.md
(or wherever it was created):
| <date> | <agent-id> | <gpt-name> | https://github.com/linexjlin/GPTs/blob/main/prompts/<filename> | distilled |
Report — Show user: agent updated, which GPT was source, what topic was added to knowledge/.