@agentwares/agentguard

MCP server · npm · graded 2026-10-20Source linked

okgate (formerly agentguard): nothing irreversible without an OK. MCP policy proxy for AI agents: hard spend limits, destructive-action gating with approvals, a kill switch, scoped credentials, dry-run writes with mutation diffs, loop breaker, blast-radiu

No outside reports on file

@agentwares/agentguard fetches arbitrary web URLs and requires API keys or tokens; restrict its folder reach, network access, and credentials before installing.

Based on public adoption and transparency signals as of Oct 19, 2026. Our opinion, not a code audit or security guarantee.

Was this verdict for @agentwares/agentguard helpful?

What it can access

Local filesNo local folder access notedLOW
Command executionNo shell command runner notedLOW
Network reachFetches arbitrary web URLsREVIEW
CredentialsRequires API key or tokenREVIEW
Package sourcenpm (@agentwares/agentguard)PUBLIC

Inferred from this listing’s public package details and description. Check the project’s source before installing.

Install

npx -y @agentwares/agentguard

All setup options ↓

Check this skill Watch it in my stack Compare Add to allowlist README badge View project

About @agentwares/agentguard

@agentwares/agentguard is an MCP server focused on General: okgate (formerly agentguard): nothing irreversible without an OK. MCP policy proxy for AI agents: hard spend limits, destructive-action gating with approvals, a kill switch, scoped credentials, dry-run writes with mutation diffs, loop breaker, blast-radiu.

It is published on npm as @agentwares/agentguard and hosted on GitHub at agentwares/agentguard. Based on its public metadata, its operational scope includes: fetches arbitrary web urls, requires api key or token.

Setup

These use the package this listing points to. Check the project’s own instructions before running anything.

npm command

npx -y @agentwares/agentguard

MCP client config

For Claude Desktop (claude_desktop_config.json) or Cursor (.cursor/mcp.json).

json

{
  "mcpServers": {
    "agentguard": {
      "command": "npx",
      "args": [
        "-y",
        "@agentwares/agentguard"
      ]
    }
  }
}

Why this grade

DD — High risk signals

Signal-based grade (1/7 points): public source code. This reflects adoption and transparency, not a code review.

Popularity and upkeep

8/100

Listing last refreshed: 2026-10-28

These are public popularity signals, not a maintenance check. Look at the repository’s commit history and open issues.

Security reports

No reports found for this package in our research (last checked 2026-10-29).

Check a specific version →

Risks to watch for

Untrusted web content & prompt injection: Fetches arbitrary web pages or remote endpoints where hidden instructions could manipulate agent behavior. Restrict outbound domains and review tool outputs.
Credential & token exposure: Requires API keys or secrets in configuration. Store credentials in environment variables with read-only or minimum required scopes and rotate them regularly.
Dependency & version pinning: Distributed via npm (@agentwares/agentguard). Pin an exact release version or commit SHA rather than floating tags so upstream updates are reviewed first.

How to lower the risk

Least-privilege permissions

Give it access only to the specific folders, accounts or projects it needs, never your whole drive or admin account.
Use a separate, limited API key or login just for this tool so you can switch it off without affecting anything else.

Network isolation

Run it on a work computer or sandbox account, not one holding personal or company-sensitive data.
If your IT team can, limit which websites and services it is allowed to contact.

Monitoring

Check its activity history or logs weekly at first, and look at what it changed or sent.
Turn on alerts for new logins or unusual usage on any account it connects to.

Red flags to watch for

It asks for more access than its job requires, or prompts you to paste passwords.
It suddenly updates with new owners, or starts contacting unfamiliar websites.

Questions

How do I install @agentwares/agentguard?

Run npx -y @agentwares/agentguard in your terminal, or add the server entry to your claude_desktop_config.json or .cursor/mcp.json as shown in the Setup section on this page. Full setup docs and environment variables are in github.com/agentwares/agentguard.

Are there security reports about @agentwares/agentguard?

No reports found for this package in our research (last checked 2026-10-29). @agentwares/agentguard currently holds Grade D (High risk signals) based on public adoption and transparency signals. Review the permissions and source repository before granting access to local files, shell commands, or API keys.

Is @agentwares/agentguard actively maintained?

We do not have star or download counts on file for @agentwares/agentguard yet (listing refreshed 2026-10-28). Check the commit history and open issues at github.com/agentwares/agentguard before relying on it.

What are alternatives to @agentwares/agentguard?

Compare @agentwares/agentguard side by side with similar MCP servers in the Related tools section on this page, or browse all MCP servers on AgentArmorHQ.

Maintain or use @agentwares/agentguard? Show its live grade in your GitHub README, docs site, or run a check from the terminal before you install.

AgentArmorHQ Risk Grade D for @agentwares/agentguard

GitHub README Markdown:

markdown

[![AgentArmorHQ Risk Grade for @agentwares/agentguard](https://agentarmorhq.com/api/badge/agentwares-agentguard)](https://agentarmorhq.com/skills/agentwares-agentguard)

HTML embed:

html

<a href="https://agentarmorhq.com/skills/agentwares-agentguard"><img src="https://agentarmorhq.com/api/badge/agentwares-agentguard" alt="AgentArmorHQ Risk Grade for @agentwares/agentguard" height="20" /></a>

Terminal check:

terminal

curl -fsSL https://agentarmorhq.com/api/cli | node --input-type=module - check "agentwares-agentguard"

Share with your security lead

Need sign-off to use @agentwares/agentguard? Copy this note for Slack, a ticket or a pull request.

### AgentArmorHQ Security Review Memo: @agentwares/agentguard
- **Listing URL:** https://agentarmorhq.com/skills/agentwares-agentguard
- **Type:** App connector
- **Risk Grade:** Grade D (High risk signals)
- **Outside Security Advisories on File:** 0

#### Permission & Blast-Radius Signals
- **Local files:** No local folder access noted
- **Command execution:** No shell command runner noted
- **Network reach:** Fetches arbitrary web URLs
- **Credentials:** Requires API key or token
- **Package source:** npm (@agentwares/agentguard)

#### Recommended Hardened Setup
- **Command:** `npx -y --ignore-scripts @agentwares/agentguard`
- **Controls:** Pin an exact version in `.mcp.json`, scope access to a single workspace folder, and use a least-privilege read-only token where supported.

_Based on public adoption and transparency signals as of Oct 19, 2026. Our opinion, not a code audit or security guarantee._

Key facts

Type: App connector
Publisher: agentwares
Source code: github.com/agentwares/agentguard
Install command: npm install @agentwares/agentguard
Est. context load: ~600 tokens / turn (~3 tool schemas, 0.3% of 200k context)
Last updated: 2026-10-28
Last graded: 2026-10-20

Community

Discussions for @agentwares/agentguard

Ask questions or share configuration tips. Governed by our Community Rules & Terms.

No posts match this filter yet.

Be the first to ask a question or share a working tip.

Was this listing for @agentwares/agentguard helpful?

Check before you install, in Claude

Your assistant answers Block, Review or No known issues for any package before it installs it.

Get this check inside Claude