Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Keychains proxies API requests and injects real credentials server-side using placeholder variables like `{{OAUTH2_ACCESS_TOKEN}}`. Agents send requests with placeholders; the Keychains proxy substitutes real tokens before forwarding to the upstream API. Users manage credentials and approve agent access from the Keychains dashboard.
Agents can call any API without ever receiving real credentials, eliminating the risk of key exposure through logs, prompts, or agent memory.
When to use it
- List GitHub repos from an AI agent without exposing OAuth tokens
- Send Slack messages via an agent without sharing bot credentials
- Read Gmail through an assistant without handing it your OAuth token
- Query Stripe customer data from an agent without embedding secret keys
- Run cloud-hosted agents that call third-party APIs without credential leakage
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: List my GitHub repositories
- 1Run `keychains curl https://api.github.com/user/repos -H 'Authorization: Bearer {{OAUTH2_ACCESS_TOKEN}}'`
- 2Receive an approval URL from Keychains on first use
- 3Present the approval URL to the user for FaceID/Passkey authentication
- 4Run `keychains wait <approval-url>` to poll until the user approves
- 5Replay the original request and receive the live API response
JSON list of the user's GitHub repositories, with no credentials ever passed to the agent
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.