Turnkey Agent Skills

A collection of AI agent skills for Turnkey — the wallet infrastructure platform that manages cryptographic keys in hardware-backed secure enclaves.

These skills enable AI agents to autonomously create wallets, derive addresses, sign transactions, and authenticate users across multiple blockchains. Compatible with Claude Code, OpenAI, and other third-party agent frameworks. Use of these Skills is subject to Turnkey's Terms of Service. See also: Shared Responsibility Model.

What is Turnkey?

Turnkey stores private keys in secure enclaves (AWS Nitro, etc.) — keys are never exposed to application code. Every API request is cryptographically "stamped" (signed with your API key pair) before being sent, ensuring only authorized callers can trigger operations on your keys.

For full documentation, please see our docs site or our docs MCP server.

Why Use These Skills?

These skills let you operate Turnkey through conversation instead of code. Point your AI agent at the skills and describe what you need — it handles the API calls, parameter formatting, and chain-specific details.

• Explore and test — create wallets, sign transactions, and set policies without writing integration code. See managing-wallets and signing-transactions.
• Administer your organization — manage users, rotate API keys, and monitor activities conversationally. See managing-users, managing-policies, and monitoring-activities.
• Provision autonomous agents — set up a scoped wallet with constrained credentials and governance policies so an agent can transact on-chain without human review. See provisioning-agent.

Environment Variables

All skills require these three variables:

TURNKEY_API_PUBLIC_KEY=<your-api-public-key>
TURNKEY_API_PRIVATE_KEY=<your-api-private-key>
TURNKEY_ORGANIZATION_ID=<your-organization-id>

⚠️ Security warning — a note on root credentials

The credentials you provide here may belong to a root user or a scoped non-root user. Root users are members of your organization's root quorum — they can execute any action and bypass all policies. Before giving these credentials to an AI agent, consider your use case:

• Interactive assistant (human approves each action): root credentials may be acceptable for testing — evaluate the risk before using them in production.
• Autonomous agent (acts without human review): do not use root credentials. Create scoped credentials with policies that limit what the agent can do — see skills/provisioning-agent/.

LLMs can misinterpret instructions or execute unintended actions. Scoped credentials ensure mistakes are bounded. Turnkey policies are the technical enforcement mechanism.

Get these from the Turnkey console under Settings → API Keys. When you create an API key, you receive a P-256 public/private key pair. The organization ID is visible in the URL and settings page.

Loading Skills

Each skill is a SKILL.md file — a structured prompt that teaches an AI agent how to perform a Turnkey operation. The root SKILL.md is the master index: it lists every skill, explains when to use each one, and defines the load order for multi-step tasks.

Clone the repo and point your assistant at the folder. This works with any AI assistant that can read files (Cursor, Cline, Windsurf, Aider, ChatGPT with file upload, etc.):

git clone https://github.com/tkhq/turnkey-agent-skills.git
cd turnkey-agent-skills

Then reference the root SKILL.md (or a specific skill) in your prompt:

Please read SKILL.md in this repo and help me sign an Ethereum transaction.

Skill load order

For multi-step tasks, skills should be loaded in a specific order. The root SKILL.md documents this in detail, but the summary is:

• First-time setup — getting-started (verifies credentials, creates your first wallet)
• Signing a transaction — managing-wallets → signing-transactions (one skill covers EVM, Solana, Bitcoin, and 10+ other chains)
• Provisioning an agent — provisioning-agent (pulls in managing-wallets, managing-users, and managing-policies as needed)

Agents that load the root SKILL.md first will follow this order automatically.

Skills

Workflows are guided multi-step procedures (start here if you're new). Primitives are individual operations (use these for ongoing work).

Workflows

Guided multi-step procedures — start here for onboarding or agent setup.

Skill	Path	Description
Getting Started	skills/getting-started/	Day-0 onboarding: verify credentials, create your first wallet
Provisioning Agent	skills/provisioning-agent/	Create a scoped agent with constrained credentials and policies
Managing Agent	skills/managing-agent/	Debug denied transactions, rotate keys, update agent policies

Primitives

Individual operations — use for ongoing work and one-off tasks.

Skill	Path	Description
Managing Wallets	skills/managing-wallets/	Create wallets, derive addresses, add chains, import/export
Signing Transactions	skills/signing-transactions/	Sign and broadcast on any chain (EVM, Solana, Bitcoin, +10 more)
Managing Users	skills/managing-users/	Create users, rotate API keys, user tags
Managing Policies	skills/managing-policies/	Access control, spending limits, allowlists, multi-sig
Monitoring Activities	skills/monitoring-activities/	Activity status, consensus approvals, audit

Note: You don't need to run npm install in this repo to use the skills — your AI assistant reads the SKILL.md files directly as prompts. The skills will, however, tell you to install @turnkey/sdk-server (and a chain-specific package if you're signing transactions) in your own project, since Turnkey's API requires a cryptographic stamp on every request that the SDK handles for you. The Running Examples section below is only for developers who want to run this repo's examples/ demos end-to-end.

Running Examples

# Install dependencies
npm install

# Copy and fill in environment variables
cp .env.example .env
# Edit .env with your API_PUBLIC_KEY, API_PRIVATE_KEY, ORGANIZATION_ID

# Run an example (requires tsx)
npx tsx examples/wallet-management.ts
npx tsx examples/ethereum-ethers.ts
npx tsx examples/ethereum-viem.ts
npx tsx examples/solana-signing.ts
npx tsx examples/bitcoin-signing.ts

Versioning

This repo is versioned as a single bundle using Semantic Versioning. The authoritative version lives in package.json; release-please keeps .claude-plugin/plugin.json in sync via its extra-files configuration. See CHANGELOG.md for release history.

SemVer contract for this skill bundle:

• Major — removing or renaming a skill, removing a documented workflow step, or changing required environment variables.
• Minor — adding a new skill, adding a new optional reference, or non-breaking expansions to an existing skill's workflow.
• Patch — wording, formatting, doc fixes, dependency bumps with no behavior change.

Pinning a release. Each distribution channel resolves to a git tag, so you can pin to a specific version. Replace <version> with a tag from the GitHub releases page:

# Local clone
git clone --branch v<version> https://github.com/tkhq/turnkey-agent-skills.git

Releases are automated via release-please on merges to main. Use Conventional Commits (feat:, fix:, feat!: for breaking) so the bot can compute the next version.

Contributing

Adding a new skill, running the eval harness, or exploring the project structure? See CONTRIBUTING.md.