The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill suggests installing the zentao-cli package from the official npm registry. This is a vendor-owned resource (Easysoft) and is consistent with the skill's purpose.
[REMOTE_CODE_EXECUTION]: The documentation includes the use of npx zentao-cli, which downloads and executes the latest version of the tool. As this targets the official vendor package on a well-known service, it is considered safe runtime behavior.
[COMMAND_EXECUTION]: The skill executes the zentao command-line utility to manage project data. It explicitly includes a --yes flag for deletion operations to facilitate autonomous agent workflows.
[CREDENTIALS_UNSAFE]: The skill documentation describes credential storage locations such as ~/.config/zentao/zentao.json and environment variables. Crucially, the instructions provide safety guidelines strictly forbidding the agent from reading these sensitive files or collecting passwords from users directly.
[PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it ingests and displays data from external ZenTao records.
Ingestion points: Data returned from the ZenTao API via commands like zentao bug or zentao story (SKILL.md).
Boundary markers: Absent; instructions do not specify delimiters for external content in the agent's output.
Capability inventory: The agent can execute a wide range of CLI commands including creating, updating, and deleting data using the zentao tool.
Sanitization: No explicit sanitization or filtering of external content is mentioned in the instructions.

zentao-cli