The Agent Skills Directory

[SAFE]: The skill provides comprehensive guidance on the secure management of CloudBase Run environments, emphasizing the use of the manageAppAuth tool to generate environment-specific API keys instead of reusing local client credentials.\n- [SAFE]: The instructions include specific warnings against common security pitfalls, such as echoing sensitive environment variables (e.g., CLOUDBASE_APIKEY) or temporary platform credentials (x-cloudbase-context) in HTTP responses.\n- [SAFE]: Deployment workflows for container images and source code utilize platform-native tools (manageCloudRun) and follow industry-standard practices, including the use of VPC configurations to secure database connectivity.\n- [SAFE]: The troubleshooting documentation provides a structured approach to diagnosing deployment issues through official runtime logs (getProcessLog), explicitly advising against arbitrary changes to security-sensitive parameters like readiness probe delays without evidence.\n- [INDIRECT_PROMPT_INJECTION]: The skill manages a deployment attack surface where user-provided code or images are processed. This is mitigated by clear boundary markers (initialization checks), capability inventory (restricted to manageCloudRun), and sanitization (credential decision gates).

cloudrun-development