hush-uam

Hush Agent Skills

Plugins, skills, and rules for AI coding agents (Claude Code, and future support for other tools) that help you work with Hush Security products.

Plugins

Plugin	Description
hush-uam	Author Kubernetes manifests for the Hush UAM operator — AccessCredential, AccessPrivilege, AccessPolicy CRDs under am.hush.security/v1alpha1.

Installation

Claude Code (recommended)

Add this repo as a marketplace and install the plugin:

/plugin marketplace add hushsecurity/agent-skills
/plugin install hush-uam

Or install a specific skill manually by cloning into your skills directory:

# user-level (available in all projects)
git clone https://github.com/hushsecurity/agent-skills.git ~/hush-agent-skills
ln -s ~/hush-agent-skills/plugins/hush-uam/skills/hush-uam-manifest ~/.claude/skills/hush-uam-manifest

# or project-level (checked into the repo)
git clone https://github.com/hushsecurity/agent-skills.git /tmp/hush-agent-skills
cp -r /tmp/hush-agent-skills/plugins/hush-uam/skills/hush-uam-manifest \
      <your-project>/.claude/skills/

Cursor

A pre-built Cursor project-rule wrapper lives at tools/cursor/. Copy it into your project's .cursor/rules/ directory:

git clone https://github.com/hushsecurity/agent-skills.git /tmp/agent-skills
mkdir -p <your-project>/.cursor/rules
cp -r /tmp/agent-skills/tools/cursor/hush-uam <your-project>/.cursor/rules/

See tools/cursor/README.md for capability caveats (Cursor doesn't have Claude Code's structured-question tool — questions go out as prose).

Other AI tools

Wrappers for additional non-Claude-Code agents live under tools/ — see tools/README.md for the current list and contribution notes. If your tool isn't listed yet, the skill content under plugins/hush-uam/skills/hush-uam-manifest/ is plain Markdown that any LLM-driven editor can read; consult your tool's docs on how to point it at a custom rule/instruction file.

Usage

Once installed in Claude Code, ask things like:

• "Create an access policy for postgres"
• "I need a Hush credential for our Gemini project"
• "Generate manifests for an OpenAI policy bound to my staging namespace"

The skill activates automatically based on intent and walks you through the necessary inputs (scope, attestation, delivery, credential fields, secret strategy, privileges) using structured questions.

Repository layout

.
├── plugins/                              ← Claude Code plugins (canonical content)
│   └── hush-uam/
│       ├── .claude-plugin/plugin.json    ← plugin manifest
│       └── skills/
│           └── hush-uam-manifest/
│               ├── SKILL.md
│               └── references/           ← portable Markdown reference docs
├── .claude-plugin/marketplace.json       ← marketplace manifest (this repo as a Claude Code marketplace)
└── tools/                                ← wrappers for non-Claude-Code AI tools (Cursor, etc.)

License

Apache 2.0 — see LICENSE.