The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes tools that execute shell commands defined in a project's local configuration.
Evidence: The deployBuild and deployApply tools execute buildCommand, installCommand, and deployCmd declarations found in cloudbaserc files.
Mitigation: The skill implements a 'Plan before apply' workflow and enforces a requirement for confirm=true and confirmDestructive=true parameters before performing write operations.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted configuration data that influences agent actions and command execution.
Ingestion points: cloudbaserc.json, cloudbaserc.yaml, cloudbaserc.yml, and cloudbaserc.js files located in the project root.
Capability inventory: Shell command execution, local file system reads, and cloud resource management via the deployBuild, deployPlan, and deployApply tools.
Boundary markers: The skill performs schema validation against the official JSON schema before execution.
Sanitization: Uses template variable rendering ({{env.*}}) for environment-specific data and encourages storing secrets in .env files rather than the config itself.
[EXTERNAL_DOWNLOADS]: The skill references an external resource for configuration validation.
Evidence: Fetches the official JSON schema from https://static.cloudbase.net/cli/cloudbaserc.schema.json to validate project configurations. This is an official resource from the vendor 'tencentcloudbase'.

cloudbase-declarative-deploy