A local security check for repositories that may be given to an AI agent, committed, or shared. It looks for exposed API keys, MCP servers that run unpinned code, and credential files missing from .gitignore; MCP servers are extensions that let an agent use outside tools.

It helps decide whether a repository is safe to run with an AI agent and safe to commit or share.

It helps catch secrets and risky agent configuration before they leave the developer's machine or enter version control. The repository is checked locally rather than uploaded.

holster — Plugin by nauta-ai

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Claude Code

/plugin marketplace add nauta-ai/holster

agentmods

npx agentmods add plugins/nauta-ai/holster/holster

Clone the repo

git clone --depth 1 https://github.com/nauta-ai/holster

Made for: Claude Code.

Per session not measured What this adds to a session before it is invoked.

When invoked not measured Not applicable: nothing here is loaded into a session.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Security

Grade A, and why

holster scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/holster/.claude-plugin/plugin.json · 12 lines

What it actually says

{
  "name": "holster",
  "description": "Preflight security for AI-agent development. Run `holster guard .` to get a RUN verdict and a SHARE verdict on a repo before you hand it to an AI agent, commit it, or share it — catches live API keys, MCP servers that run unpinned code, and credential files missing from .gitignore. Local-first; the repo never leaves your machine. Not a password manager.",
  "version": "0.1.0",
  "mcpServers": {
    "holster": {
      "command": "uvx",
      "args": ["holster-mcp"]
    }
  }
}

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 12 lines scan A 4f9f2c491487

Subscribe to this mod's changes

holster is a plugin published in the GitHub repository nauta-ai/holster (0 stars, last pushed 3mo ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Security ai-agents claude-code codex cursor holster local-first

pad

Talk to your project. Pad is local-first project management for developers and AI agents — this plugin gives Claude Code a conversational Pad skill, typed shortcuts (/pad:status, /pad:capture, /pad:onboard, /pad:connect, /pad:disconnect), and consent-gated live item-change notifications.

PerpetualSoftware/pad · not measured

sverklo-skill

Teaches Claude how to use sverklo's code intelligence MCP tools — semantic search, blast-radius, PR review, codebase audit, persistent memory.

sverklo/sverklo · not measured

lians

Bitemporal memory for Claude Code with deterministic supersession, point-in-time recall, local MCP, erasure proofs, and tamper-evident history.

Lians-ai/Lians · not measured

piia-engram

Local-first personal AI identity layer for MCP-compatible coding tools. Stores your lessons, decisions, playbooks, and project context as local JSON. AI proposes; high-risk items wait for your review, and everything stays visible and reversible. Local-first, no cloud, no account.

Patdolitse/piia-engram · not measured

mnema

Puts the project's committed mnema record into the session's opening context without the agent asking, and connects the mnema MCP server so it can record its own work.

felipesauer/mnema · not measured

palimem marketplace

Plugin marketplace listing 1 plugin: memory-service.

palimem/palimem · not measured