The Agent Skills Directory

[SAFE]: The skill is authored by tencentcloudbase and correctly references legitimate vendor resources, including the Tencent Cloud Container Registry (ccr.ccs.tencentyun.com) and the tcbr API namespace.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill includes defensive instructions to protect sensitive platform credentials. Specifically, it warns agents not to echo the x-cloudbase-context header (which contains temporary cloud access credentials) or return credential-related environment variables in HTTP responses.
[INDIRECT_PROMPT_INJECTION]: The skill manages a deployment surface that involves user-provided source code and container images. However, it mitigates these risks by requiring compliance with established security protocols, such as sensitive-runtime-data-protection.md and deployment-gate.md, and by providing clear checklists for validating external image configurations. Ingestion points: manageCloudRun(action="deploy") parameters targetPath and imageUrl (SKILL.md). Boundary markers: Mandatory use of sensitive-runtime-data-protection.md and deployment-gate.md protocols (SKILL.md). Capability inventory: manageCloudRun (deploy, update, traffic) and queryCloudRun (detail, logs) (SKILL.md). Sanitization: Explicit constraints on echoing headers or env vars in handlers (SKILL.md).
[COMMAND_EXECUTION]: All command execution patterns associated with the manageCloudRun and queryCloudRun tools are consistent with their intended administrative purpose and are used with appropriate safety guidance, such as requiring absolute paths for target directories.

cloudrun-development