Deploy apps with AI tools
GrStack is also known as Stackr. CLI commands and integration URLs in this guide use the Stackr name.
Connect a project in your Stackr workspace, stage a release, and verify the preview before it serves live traffic.
Connect Codex, Kiro, or Gemini CLI
- Open Console in your Stackr workspace and select a project.
- Create an Agent token and store it as
STACKR_AGENT_TOKENin your client’s environment. - Use the setup snippet for your client. Set
PROJECT_IDto your Stackr project ID.
npm install -g https://stackr.grasany.com/grasany-stackr-cli-0.13.0.tgz
stackr connect codex PROJECT_ID
Codex
Use the output from stackr connect codex PROJECT_ID in your Codex config.toml. Restart Codex with the token in its environment. Run codex mcp list or open /mcp to check the connection.
Kiro
Use the output from stackr connect kiro PROJECT_ID in Kiro’s MCP configuration. Approve STACKR_AGENT_TOKEN for environment expansion when Kiro asks.
Gemini CLI
Use the output from stackr connect gemini PROJECT_ID in your Gemini CLI settings.json. Start Gemini CLI with the token in its environment and run /mcp to check the connection.
Agents can inspect usage, list releases, check release status, and stage managed starters. You can revoke the Agent token in Console.
Use ChatGPT or Gemini on the web
Where custom MCP connections are available, add this public endpoint to get Stackr deployment guidance.
https://stackr.grasany.com/api/mcp
Choose no authentication when the client asks. This endpoint returns public guidance. Use your terminal or a connected IDE with a project token to work with private releases. You can also give a chat assistant the Stackr guide for AI tools and ask it to prepare a build.
Official OpenAI documentation for ChatGPT custom MCP · Gemini custom connected apps
Stage, verify, and publish
Create a deploy token in your project’s deployment history. Put the token and endpoint in your terminal or CI secret store as STACKR_DEPLOY_TOKEN and STACKR_DEPLOY_URL.
stackr stage launch.json
stackr launch dist
stackr preview RELEASE_ID
stackr smoke RELEASE_ID --contains "Expected text"
stackr releases
stackr promote RELEASE_ID
Build the app locally or in CI first, then use stackr launch dist for a static output directory. Use stackr deploy-static site.zip if you already have a ZIP. The CLI smoke test checks the preview HTTP response and optional expected text. Open the preview in a browser to test scripts and interactions. Publish a reviewed release with stackr promote or from the dashboard. To roll back, promote a known good release from history.