skill-security
Medium Risk 71Audit AI Agent Skills Before Installation
Unvetted agent skills can hide prompt injection, credential theft, persistence, or unsafe dependencies. This skill combines offline scanning with guided semantic review.
Unified search
Start with the task. Skillstore will show complete packs first, then individual skills when they are a better match.
Showing results for "agent"
Use these when you need one narrow capability instead of a whole pack.
Audit AI Agent Skills Before Installation
Unvetted agent skills can hide prompt injection, credential theft, persistence, or unsafe dependencies. This skill combines offline scanning with guided semantic review.
Build WeChat Mini Program Auth with CloudBase
WeChat Mini Program auth is easy to misapply when CloudBase identity is automatic. This skill gives practical patterns for wx.cloud, OPENID, UNIONID, and cloud functions.
Manage CloudBase MySQL with MCP
CloudBase MySQL operations need clear tool choice and safety checks. This skill guides agents through MCP-based provisioning, querying, schema changes, and permission reviews.
Preserve Project Context Across AI Sessions
AI coding sessions lose important project context over time. This skill keeps decisions, current work, and progress in structured Markdown files.
Compare Parallel Implementations
Complex features often have several valid implementation paths. This skill runs independent agent implementations, reviews them, and helps select the strongest result.
Validate Features with Scenario Tests
Feature work can pass shallow tests while failing against real systems. This skill guides Claude, Codex, and Claude Code to prove behavior with scenario tests and documented real-dependency checks.
Compare Parallel Implementations
Complex software choices can be hard to judge from discussion alone. Test Kitchen guides parallel designs, implementations, reviews, and winner selection.
Orchestrate Parallel Coding Agents
Large features are hard to split across autonomous coding agents. This skill provides a worktree-based workflow for issue design, wave planning, monitoring, merging, and cleanup.
Manage Multi-Agent Context
Long-running AI work often loses decisions, blockers, and session state. This skill provides practical workflows for checkpoints, handoffs, memory records, and compact context sharing.
Create and Edit Music with ACE Step
Music generation and repair can require separate tools and costly services. This skill routes generation, inpainting, and outpainting through ACE Step on RunComfy.
Generate Music with ElevenLabs on RunComfy
Creating structured music from text often requires specialized audio tools and manual setup. This skill guides RunComfy CLI requests for songs, instrumentals, jingles, and loops.
Inpaint Images with RunComfy
Precise image edits are hard when only one region should change. This skill routes Claude, Codex, or Claude Code to RunComfy inpainting models with mask-aware prompts and safe CLI usage.