Command Code + Raims
A terminal agent for planning a Raims integration from your application code.
Already using this agent? Go to project context and keep your existing setup.
Quick start
Prepare the agent
Check the requirements and install the agent using the official quickstart. The npm installation uses:
Shellnpm i -g command-code@latestOpen your project
In your application folder, run the command below and complete the agent's sign-in. On native Windows, the short alias is
cmdc.cmdccommand-codeAdd the documentation
Ask the agent to read Raims' documentation index and open the relevant guides. llms-full.txt provides the consolidated content. Start with authentication and charges.
Try a prompt
Read https://docs.raims.com/llms.txt and the authentication and charges guides. Inspect this application's backend and propose a plan to retrieve an existing Sandbox charge with GET /v1/charges/{id}. Prepare an example using protected server configuration for the key. Do not send requests or use real credentials. Show the sources and the files you would change before editing.
Project context
Add the content below to AGENTS.md at the application root. Command Code uses this file as project memory. The /memory command lets you open the file for editing within the session.
The instruction file belongs with your project. This tree shows its location:
# Raims integration
- Read https://docs.raims.com/llms.txt before choosing endpoints and fields.
- Use https://docs.raims.com/llms-full.txt for the consolidated documentation.
- API base URL: https://api.raims.com.
- In Raims product installation examples, select the latest stable version
through the package manager (`latest` where supported).
- Prepare examples for Sandbox, with Bearer authentication in the backend.
- Keep the key in protected server configuration. Never put secrets in code,
prompts, logs, or browser applications.
- Explain the plan and request authorization before sending API requests.Verify the setup
Ask the agent to identify the guides it read, the API base URL, and the place in your backend where the key will be loaded. It should prepare the plan and example without asking you to paste a real credential.
Reading the documentation does not authenticate the application. Requests use a Raims API key for the intended company and environment.
Recognize a completed setup
A reviewable plan to retrieve an existing Sandbox charge, with public sources and proposed backend files. The prompt asks for planning only, with no API request and no real credential in the chat.
| Check | What the plan should show |
|---|---|
| Project instructions | The agent considered AGENTS.md and kept the application's existing conventions. |
| Sources | Authentication and charges guides, with the links it actually read. |
| Environment and endpoint | Sandbox and GET /v1/charges/{id}, using an ID from your integration. |
| Secret | A protected backend configuration point, without a key in code or prompts. |
| Proposed work | The files to change and the result to inspect before a test is authorized. |
If an item is missing, ask the agent to complete the plan before editing or authorizing a request. If it could not read a source, provide the relevant public excerpt and keep the missing information explicit.
Common issues
Use command-code or the cmdc alias on native Windows. The name cmd
also identifies the Windows command processor.
Open the relevant guide in your browser and provide the public excerpt as context. Ask the agent to identify any missing information before suggesting code.
Ask the agent to review the authentication guide. The backend sends the Raims key with Bearer authentication; an AI provider credential does not replace it.