feat: add HOL Guard security plugin

.factory-plugin/marketplace.json

@@ -12,6 +12,12 @@
       "source": "./plugins/security-engineer",
       "category": "security"
     },
+    {
+      "name": "hol-guard",
+      "description": "HOL Guard security workflows for supported AI harness protection, approval evidence, and pre-trust scanning",
+      "source": "./plugins/hol-guard",
+      "category": "security"
+    },
     {
       "name": "droid-evolved",
       "description": "Skills for continuous learning and improvement: session navigation, human writing, skill creation, visual design, and browser automation",

plugins/hol-guard/.factory-plugin/plugin.json

@@ -0,0 +1,5 @@
+{
+  "name": "hol-guard",
+  "description": "HOL Guard security workflows for supported AI harness protection, approval evidence, and pre-trust scanning",
+  "version": "1.0.0"
+}

plugins/hol-guard/skills/hol-guard/SKILL.md

@@ -0,0 +1,126 @@
+---
+name: hol-guard
+description: Set up HOL Guard for supported coding-agent harnesses, review Guard approvals and receipts, or scan skills, plugins, MCP servers, and agent packages before trust.
+version: 1.0.0
+tags: [security, agent-security, mcp, supply-chain, approvals]
+---
+
+# HOL Guard
+
+Use HOL Guard when a user wants local AI-agent security for a supported harness, needs to review Guard approvals/evidence, or wants to scan an agent package before installing or trusting it.
+
+HOL Guard is an external runtime. This Factory plugin does not claim that Droid itself is a HOL Guard-protected harness and does not replace Factory's own permissions or security controls.
+
+## Hard Rules
+
+- Never read `.env` files or print secrets to diagnose Guard.
+- Never bypass a Guard block or unresolved approval.
+- Do not claim protection until `hol-guard status` or a harness-specific doctor command proves it.
+- Treat `hol-guard command test` as inspection only, not an enforcement wrapper.
+- Keep Guard Cloud optional. Local protection and scanning do not require a Cloud connection.
+- Preserve user configuration; prefer Guard-owned setup commands over manual harness config edits.
+
+## Install
+
+Check the runtime and scanner separately:
+
+```bash
+command -v hol-guard || true
+command -v plugin-scanner || true
+```
+
+For runtime protection:
+
+```bash
+pipx install hol-guard
+hol-guard status
+hol-guard detect --json
+```
+
+For package or skill scanning:
+
+```bash
+pipx install plugin-scanner
+plugin-scanner lint .
+plugin-scanner verify .
+```
+
+Do not assume the `hol-guard` package also provides the `plugin-scanner` command.
+
+## Protect A Supported Harness
+
+HOL Guard currently supports these harness targets: `codex`, `claude-code`, `copilot`, `cursor`, `gemini`, `hermes`, `openclaw`, `opencode`, and `antigravity`.
+
+Use Guard-owned setup and verify before normal work:
+
+```bash
+hol-guard bootstrap
+hol-guard install <harness>
+hol-guard run <harness> --dry-run
+hol-guard run <harness>
+hol-guard doctor <harness> --json
+hol-guard status
+```
+
+For Hermes, prefer its dedicated bootstrap path when applicable:
+
+```bash
+hol-guard hermes bootstrap
+```
+
+If the current environment is Factory/Droid without one of the supported harnesses above, do not invent a `droid` target. Use HOL Guard for package scanning, evidence review, or for a separately installed supported harness.
+
+## Review Approvals And Evidence
+
+When Guard blocks or queues work:
+
+```bash
+hol-guard approvals
+hol-guard approvals open
+hol-guard receipts
+hol-guard diff <harness>
+```
+
+Only approve after the user understands the risk reason and requested scope:
+
+```bash
+hol-guard approvals approve <request-id>
+hol-guard approvals deny <request-id>
+```
+
+For audit/evidence work:
+
+```bash
+hol-guard receipts
+hol-guard inventory
+hol-guard abom --format json
+hol-guard events
+hol-guard explain <artifact-id>
+```
+
+## Scan Skills, Plugins, MCP Servers, Or Agent Packages
+
+Run the scanner against the package root without executing the target just to scan it:
+
+```bash
+plugin-scanner lint <path>
+plugin-scanner verify <path>
+```
+
+Use the repository root for mixed agent workspaces or marketplaces so local skills, plugins, MCP configuration, and harness artifacts can be discovered together.
+
+If verification reports findings, surface the finding, affected path, and exact next safe action. Do not weaken scanner rules merely to make a package pass.
+
+## Troubleshooting
+
+```bash
+hol-guard doctor
+hol-guard detect --json
+hol-guard settings show
+plugin-scanner verify . --json
+```
+
+Report what command ran, what Guard found, what remains blocked or risky, and what proof exists. Do not claim approval, protection, or release readiness without command output proving it.
+
+Canonical project: https://hol.org/guard
+Source: https://github.com/hashgraph-online/hol-guard

skills/hol-guard

@@ -0,0 +1 @@
+../plugins/hol-guard/skills/hol-guard