[ Developers ]

Build with 1Claw

1Claw keeps API keys and credentials in HSM-backed vaults, so your AI agents can use them without ever holding them. Here's where to find everything you need to build a contest project. Projects earn the most points of any submission.

Read the quickstart ↗1claw.co/for-ai ↗docs.1claw.co ↗

[ 01 · Start in five minutes ]

  1. 01

    Create a free account

    The same 1Claw account you use to sign in here. The free tier covers 3 vaults, 50 secrets, 2 agents and 1,000 API calls a month, no card needed.

  2. 02

    Run the setup

    Finds Cursor, Claude Code, Windsurf, VS Code and other MCP clients on your machine and writes the config for each. It also creates an agent, a vault and a policy.

    npx @1claw/cli setup

  3. 03

    Check it works

    Ask your assistant to list your 1Claw secrets. You should see examples/hello.

  4. 04

    Building an agent outside your editor?

    Enroll it with a pairing code. You approve it in the browser and it gets its own API key and scoped policy.

    npx @1claw/cli agent enroll my-agent --pair

[ 02 · Let your AI assistant do the reading ]

1Claw publishes its docs in formats AI assistants read well. Paste this into Claude, Cursor, ChatGPT or any coding agent to get started.

I'm building a project with 1Claw (secure secrets and credentials for AI agents) for the 1Claw Creator Contest.

Read these first:
- https://1claw.co/llms.txt (overview, links and tier limits)
- https://1claw.co/for-ai (MCP config, agent enrollment, SKILL.md)
- https://1claw.co/skill.md (how to use the 1Claw tools)

Then help me set it up with `npx @1claw/cli setup` and build a small agent that uses a secret from my 1Claw vault without ever putting the key in code or in your context. Stick to what the free tier includes unless I say otherwise.

llms.txt ↗

Short overview for AI assistants: what 1Claw does, links and what each tier includes.

llms-full.txt ↗

The full docs in one file, for deep context.

SKILL.md ↗

Agent skill for storing, fetching, rotating and sharing secrets through 1Claw.

OpenAPI spec ↗

Machine-readable API schema for code generation and tool binding.

MCP server ↗

The tools your assistant gets once setup runs. The config lives at 1claw.co/for-ai.

[ 03 · Docs by what you're building ]

Start here

Editors and MCP

SDKs and frameworks

Going further

Want working code to start from? Clone 1clawAI/1claw-examples or browse everything on 1claw.co/developers.

[ 04 · Project ideas ]

  • A Claude Code or Cursor workflow where the agent calls a paid API (Stripe, OpenAI, GitHub) without ever seeing the key.
  • A GitHub Action that deploys or runs an agent using vault secrets through OIDC, with nothing stored in repo secrets.
  • A LangChain, CrewAI or AI SDK agent that fetches credentials just in time and logs every access.
  • A trading or payments bot that signs through the Intents API with spend caps, instead of keeping a hot wallet key in an env var.
  • A team setup where each agent has its own identity and policy, and you can revoke one without rotating shared secrets.
  • A starter template or tutorial other developers can clone to get going with 1Claw in a few minutes.

[ 05 · Make it count ]

  • Use 1Claw for real: secrets, agents, MCP, the SDK or the Intents API should be part of how it works.
  • Make it public. A GitHub repo, a live demo or a package page all work.
  • Mention 1Claw in the README or on the demo page, and explain how you use it.
  • Add your GitHub username in your profile so we can match the repo to you.
  • Never commit secrets. Keeping keys out of the code is the whole point, and the reviewers will look.

Built something?

Submit it as a project from your dashboard, then post about it for extra points.

Submit a project