shieldPRE-INSTALL SKILL DEFENSE GATEWAY

AI Skill Shield

Scan AI agent skills, SKILL.md packages, and GitHub repositories for security vulnerabilities, prompt injection, and install risks before runtime execution.

Explore public GitHub scans →Try a sample report

link

Paste a public GitHub or skills.sh URL. Full imports include install scripts, workflows, registries, and submodules before validation. No account is required.

Target: remote_repositoryEngine: v2.0.0

QUICK START INTELLIGENCE

verified_user

11 Validation Axes

Comprehensive multidimensional analysis covering frontmatter, AST code, permissions, and supply chain.

warning

12 Threat Categories

Real-time heuristics detecting prompt injection, credential exposure, obfuscation, and dangerous shell pipelines.

terminal

23 Agent Signatures

Runtime markers for Claude, Codex, OpenClaw, Cursor, OpenAI Agents, and LangChain ecosystem tools.

What AI Skill Shield does

cloud_upload

Upload

Drop a local skill package or scan the files directly

travel_explore

Audit

Review GitHub install scripts, workflows, registries, and runtime risk

description

Report

Inspect score, findings, repo audit evidence, and export artifacts

What is AI Skill Shield?

AI Skill Shield is an AI agent skill security scanner, validator, and checker. It helps developers and teams inspect a skill before installation by combining static validation, repository evidence, compatibility checks, and a clear install-risk report.

It is designed for Agent Skills, SKILL.md packages, MCP-adjacent workflows, and public GitHub repositories used by AI agents.

How the security check works

1. Import: provide a repository, uploaded package, or SKILL.md.
2. Inspect: detect security findings, permissions, compatibility, and repository risks.
3. Decide: review the score, evidence, install recommendation, and exportable report.

Use the free AI skill checker →Validate a SKILL.md file →Explore security rules →Read the API docs →

AI agent skill security guides

Learn how to validate SKILL.md files and review skills before adding them to your AI-agent workflow.

SKILL.md validator

What to validate before an agent follows a skill file.

Claude Code skill security

A practical review workflow for third-party Claude Code skills.

OpenClaw skill security

How to assess OpenClaw skills before installation.

Frequently asked questions

What does AI Skill Shield check?: AI Skill Shield reviews AI agent skill files for exposed secrets, destructive commands, shell-execution risk, external network access, permission concerns, compatibility, and unsafe install-time behavior.
How can I scan an AI agent skill?: Paste a public GitHub repository URL, upload the skill files, or paste the SKILL.md content. AI Skill Shield returns a static score, highest finding severity, evidence, and an install recommendation.
Does AI Skill Shield support repository audits?: Yes. GitHub imports can inspect install scripts, workflows, registries, submodules, and repository metadata in addition to the skill files.
Can AI Skill Shield be used in automated workflows?: Yes. The AI Skill Shield API supports validation, repository auditing, policy checks, comparisons, and report exports for development and CI workflows.

QUICK START INTELLIGENCE

verified_user

11 Validation Axes

Comprehensive multidimensional analysis covering frontmatter, AST code, permissions, and supply chain.

warning

12 Threat Categories

Real-time heuristics detecting prompt injection, credential exposure, obfuscation, and dangerous shell pipelines.

terminal

23 Agent Signatures

Runtime markers for Claude, Codex, OpenClaw, Cursor, OpenAI Agents, and LangChain ecosystem tools.

What is AI Skill Shield?

It is designed for Agent Skills, SKILL.md packages, MCP-adjacent workflows, and public GitHub repositories used by AI agents.

How the security check works

1. Import: provide a repository, uploaded package, or SKILL.md.

2. Inspect: detect security findings, permissions, compatibility, and repository risks.

3. Decide: review the score, evidence, install recommendation, and exportable report.

Frequently asked questions

What does AI Skill Shield check?

AI Skill Shield reviews AI agent skill files for exposed secrets, destructive commands, shell-execution risk, external network access, permission concerns, compatibility, and unsafe install-time behavior.

How can I scan an AI agent skill?

Paste a public GitHub repository URL, upload the skill files, or paste the SKILL.md content. AI Skill Shield returns a static score, highest finding severity, evidence, and an install recommendation.

Does AI Skill Shield support repository audits?

Yes. GitHub imports can inspect install scripts, workflows, registries, submodules, and repository metadata in addition to the skill files.

Can AI Skill Shield be used in automated workflows?

Yes. The AI Skill Shield API supports validation, repository auditing, policy checks, comparisons, and report exports for development and CI workflows.