qingxuantang/tar-engine
Audit AI skill safety before you ship. Static + semantic + adversarial audit pipeline for SKILL.md, Codex skill.yaml, Claude Code commands, and OpenCode. Free hosted tier on tarai.dev.
Audit AI skill safety before you ship. Static + semantic + adversarial audit pipeline for SKILL.md, Codex skill.yaml, Claude Code commands, and OpenCode. Free hosted tier on tarai.dev.
npx add-skill qingxuantang/tar-engineAudit AI skill safety before you ship. Static + semantic + adversarial audit pipeline for SKILL.md, Codex skill.yaml, Claude Code commands, and OpenCode. Free hosted tier on tarai.dev.
Baseline-first AI agent workflow layer for ML/DL/RAG projects: context, memory, guardrails, function tools, and MCP for Codex/Claude Code.
Agent Skills for generating rich interactive HTML instead of markdown. Compatible with Claude Code, Devin, Cursor, Codex, and any Agent Skills-compatible tool.
Claude Skill that triages any task to the right model (Sonnet 4.6 / Opus 4.8 / Fable 5) and, when Fable fits, fabricates a tuned execution prompt — replies in your language. 작업을 알맞은 모델로 라우팅하고, Fable이 맞으면 실행 프롬프트까지 생성.
AI agents with support to multiple models
v3.12.1 — Persistent, auditable, contradiction-safe memory for coding agents. Hybrid BM25+vector+RRF, MIC/MAP wire format, HTTP/inbox/daemon transports, deterministic quality gates, typed lineage edges, recall explainability. 84 MCP tools, 17 AI clients, mind-mem-4b 95/95 across 10 eval categories (Qwen3.5-4B full fine-tune).
亚马逊选品 之 Listing全维度穿透分析报告 加上 全品类分析 ，关键词分析，差评分析 ，市场调研 等等。OpenClaw /claude code agent skill, amazon sorftime research MCP 智能体skill.基于 Sorftime MCP 服务和 Claude Skills 的亚马逊竞品分析工具集。