Security plugins

Keep your Claude Code development safe from malware, without thinking about it. Zero setup, free forever. Pivyon Shield checks every package Claude installs and blocks the fake and malicious ones before they can compromise your code or your machine.

2 skills

Log in and get 2FA codes through the otto vault without ever seeing passwords

1 skill · 1 MCP server

Run your app with secrets the agent never sees, ask for a missing key in the browser instead of the chat, and sync .env files with the Klef vault.

2 skills

Guardrail bench (score your hook against 33 commands, nothing executed) and a power-command deny hook with its own 31 tests.

1 skill · hooks

Find out what your AI accounts are actually sharing, then shut off the parts you never agreed to. Reads the settings pages in your own browser, checks each one against what the vendor documents, and hands you the exact click path to fix it.

1 skill · 1 command

Read-only security scan of AI agent skills, plugins and MCP configs before you install or share them

1 skill · 1 agent

Builds Gnosis Safe transactions as Forge scripts and proves the queued transaction is their output.

2 skills

Pre-push security audit: hardcoded secrets, eval/shell=True, vulnerable dependencies, unvetted network destinations.

1 skill · 1 command

Guardrail bench (score your hook against 33 commands, nothing executed) and a deny hook that reads commands like the shell, 17/17 and 16/16 on the bench, 191 built-in tests.

1 skill · hooks

Cloudflare's security-audit skill packaged for Claude Code: a source-first, coverage-led audit with isolated hunters, adversarial verifiers, a schema-validated findings.json and target-neutral reports.

1 skill · 1 agent

Scan codebases for potential personal data leaks and sensitive information

1 agent

Evaluate a GitHub repo, npm package, or X link and get a blunt INSTALL/TRIAL/SKIP/REDUNDANT/DANGEROUS verdict with evidence.

1 command

Reva AI governance for OpenAI Codex — Cedar-enforced access control for every Codex tool call and prompt. Shares the Reva PDP backend with the Claude Code plugin.

hooks

Use typed Airlock tools when a task needs authority this harness does not hold.

1 skill

Deterministic safety hooks for the portfolio: block secrets in writes/commits, verify the working tree and next migration number at session start, and require RLS on new Supabase tables.

hooks

Verify, restore and revoke the plugins your organisation signed. Hooks only — install the command first: brew install random1st/tap/skillctl

hooks

Detect, purge, and harden JS/TS repos against the 2026-Q2 wallet-stealer / supply-chain malware family.

1 skill

Connect Claude Code to the Manifest Cyber platform: assets, components, vulnerabilities, products, SBOMs, and labels as MCP tools.

1 MCP server

Blocks Claude Code from reading, changing, or deleting Google Shared drives (filesystem + Google Drive MCP connector).

hooks

Prevents accidental deletion of important files by enforcing safety checks on dotfiles, git-managed files, directories outside cwd, and special directories.

hooks

Claude Codeのセキュリティ設定を自己診断します

1 skill

Discover and vet MCP servers: search 200K+ AI tools with security verification and schemas.

1 MCP server

Authorized reverse engineering and clean-room reconstruction for websites, apps, binaries, firmware, APIs, and protocols.

1 skill

Identity security and governance for humans, non-humans and AI agents.

1 MCP server

Framework-agnostic security auditor for AI-generated and enterprise codebases. Maps findings to OWASP Top 10:2025, API Top 10, Mobile Top 10, CI/CD Top 10, LLM Top 10, and CWE.

1 skill

GRC compliance review for your pull requests. Reviews each PR against ISO 27001, SOC 2, GDPR and the EU AI Act, grounded in your company profile. Public repos are free.

1 skill

Box admin-log analytics for Claude Code: external-sharing visibility from Box's enterprise event log (admin_logs) — who shares a lot with the outside, which files get accessed from outside, and a one-shot daily_brief health check. Read-only; never revokes shares, deletes files, or mutates anything.

1 MCP server

Analyzes CVEs in RHOAI releases by querying Pyxis and Red Hat Security APIs, then triages each vulnerability to determine actual impact on OpenShift AI workloads.

1 skill

Code Security workflows for security scans, analysis, and investigation.

1 skill

비개발자도 안전하게 바이브코딩 - 위험한 작업 감지 시 친절한 설명과 함께 사용자 승인을 요청합니다.

hooks