YottaMeta

Yotta Agent Hardening — Data skill for Claude Code

Data community

How to install Yotta Agent Hardening

This entry records only its repository, not the path inside it, so there is no exact command to give. Open YottaMeta/yotta-agent-hardening and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Yotta Agent Hardening does

YuanSafe (元安全) — defensive hardening workflow for AI agents: static config-facing scan of an agent's own runtime (skills, MCP servers, tools, permissions, data surfaces) across prompt-injection / tool-call boundaries / data isolation; read-only, zero-dependency, no payloads. Part of YottaSkills.

Alternatives in Data

Privacy & Data Handling — This repository contains: - a plugin package (plugins/compound-engineering) made of markdown/config content 10.8k ★
HTML Anything — ✨ The agentic HTML editor — your local AI agent writes the HTML, you ship it 8.5k ★
Context Mode — Benchmark Results — Benchmarked against real outputs from popular Claude Code MCP servers, Skills, and dev tools 5.6k ★

README

Language: English · 中文

yotta-agent-hardening · 元安全 (YuanSafe)

YottaMeta's defensive hardening workflow for AI agents themselves: it inspects the runtime an agent lives in — installed skills, configured MCP servers, tool descriptions, permissions and data surfaces — and runs a static configuration-facing hardening scan across prompt-injection defense / tool-call boundaries / data isolation, producing a hardening report plus enforceable defense guardrails. Defense only — no attack payloads.

Triggers when the user asks for a security check / hardening of an agent or skill environment, whether MCP servers or skills can be trusted, investigation of prompt-injection / over-privilege / data-leak risks, or understanding the overall exposure after installing many skills; or says 元安全 / 加固 / 安全体检 / hardening / 扫一下我的技能 / 检查 MCP / guardrails.

Zero dependencies (Python 3.8+ standard library); Windows + Linux + macOS; defense / hardening / education oriented — no executable injection strings or attack payloads.