uni-agent
๐ฏSkillfrom zrt-ai-lab/opencode-skills
|A unified agent protocol adaptation layer that lets you call all major agent protocols (MCP, A2A, ANP, AITP) through a single API. Provides protocol routing, capability discovery, and pluggable adapters for seamless cross-protocol communication.
Same repository
zrt-ai-lab/opencode-skills(26 items)
Installation
npx vibeindex add zrt-ai-lab/opencode-skills --skill uni-agentnpx skills add zrt-ai-lab/opencode-skills --skill uni-agent~/.claude/skills/uni-agent/SKILL.mdSKILL.md
More from this repository10
ๆ้ณ่ง้ข่ชๅจๅๅธๆ่ฝใๅฝ็จๆท้่ฆๅๅธ่ง้ขๅฐๆ้ณๆถไฝฟ็จ่ฟไธชๆ่ฝใๆ่ฝๅ ๅซ๏ผ่ทๅ็ปๅฝCookieใไธไผ ่ง้ขใ่ฎพ็ฝฎๆ ้ข่ฏ้ขใๅฎๆถๅๅธ็ญๅ่ฝใ. Part of zrt-ai-lab/opencode-skills.
A subtitle generation and burn-in skill that handles the full workflow: Whisper transcription, dictionary-based correction, user review, timestamp matching, SRT generation, and FFmpeg subtitle embedding.
A skill that automates video publishing to WeChat Channels (่ง้ขๅท) using Playwright browser automation, supporting cookie-based login, title/hashtag configuration, scheduled publishing, and original content declaration.
A video editing skill that executes FFmpeg-based clip removal of filler words, verbal mistakes, and silences from recordings, then re-transcribes to verify and generates subtitle files.
An agent skill for transcribing talking-head videos and detecting speech errors using FunASR with 30-second segmented transcription, identifying stutters, filler words, and silences to generate a review transcript and deletion task list.
An environment setup skill for a Chinese video editing toolkit that installs FunASR (speech recognition with timestamps), OpenAI Whisper (subtitle generation), and FFmpeg, downloading approximately 5GB of models for speech-to-text and subtitle workflows.
A video creation skill that synthesizes images and audio into videos with fade transitions, automatic outro splicing, BGM addition, and subtitle burning, supporting a multi-stage pipeline from story scenes to final MP4.
A self-update skill for AI agents that records user feedback, traces conversation context to identify issues, and integrates lessons learned into project documentation and methodology files.
A multimodal image processing skill supporting text-to-image, image-to-image, image-to-text, and long image stitching, triggered when users mention images, generation, infographics, or OCR.
A full-dimension log analysis skill based on the RAPHL pattern that auto-identifies log types (Java App, MySQL Binlog, Nginx, Trace, Alert) and extracts 20+ entity types for root cause analysis. Supports streaming processing of 100M+ log files with low memory usage.