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.
git clone --depth 1 https://github.com/railyard-dev/railguardWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/plugins/railyard-dev/railguard/railguard)<a href="https://agentmods.dev/plugins/railyard-dev/railguard/railguard"><img src="https://agentmods.dev/badge/plugins/railyard-dev/railguard/railguard/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/plugins/railyard-dev/railguard/railguard"><img src="https://agentmods.dev/badge/plugins/railyard-dev/railguard/railguard.svg" alt="Reviewed on agentmods" width="80" height="20"></a>Grade A, and why
railguard 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 1mo 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.
What it actually says
{
"name": "railguard",
"description": "A secure runtime for Claude Code. Intercepts every tool call with policy-based allow/block/ask decisions, evasion detection, path fencing, file snapshots, and audit logging.",
"version": "0.5.1",
"author": {
"name": "railguard-dev"
},
"homepage": "https://github.com/railguard-dev/railguard",
"repository": "https://github.com/railguard-dev/railguard",
"license": "MIT",
"keywords": ["security", "guardrails", "safety", "runtime", "hooks"]
}
What it installs
The manifest is a name and a version. 3 hooks travel with it, and installing the plugin installs all of them. Each is measured on its own page, and each can be installed alone.
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.
- 1mo ago First seen · 13 lines scan A d1a054a9c3de
railguard is a plugin published in the GitHub repository railyard-dev/railguard (39 stars, last pushed 6mo 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 2026-08-30.
Other plugins, from other repositories
harness
Specification-first AI harness: 11 structural gates, 11 Ouroboros commands, 11 agent personas, and 3-tier architecture enforcement. v2.1: Pair Mode (Navigator-Driver + independent test design + /review command).
thumbgate
One 👎 becomes a hard rule the agent cannot bypass. Captures thumbs-down feedback, distills it into PreToolUse Pre-Action Checks, enforced across every future Claude Code session.
agentlint
Claude Code marketplace wrapper for AgentLint's AI-agent guardrail engine. 77 rules across 8 packs, hybrid AgentChute security feeds, circuit-breaker visibility, inline-ignore reasons.
codoop-flow
In-session orchestration for the codoop-flow Agent-Centric ticket pipeline: pick -> build -> verify -> review -> ship, driven by a skill plus a deterministic guardrail CLI.
sll
Three hooks, no agents. A stated test result must reproduce, a fabricated import never reaches disk, and every command's real output is recorded. Active in every project the moment it is installed.
auto-cc
Automatic tool-call review for Claude Code. A local encoder scores every proposed tool call before it runs, approving safe work and blocking dangerous calls with an explanation the agent can act on.