The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the zentao CLI tool to perform various operations on the ZenTao project management platform, including creating, listing, and updating records.
[EXTERNAL_DOWNLOADS]: The skill installs the @leeguoo/zentao-mcp package from the NPM registry and fetches skill configurations from the leeguooooo/zentao-mcp GitHub repository. There is a naming discrepancy between the GitHub account ('leeguooooo') and the NPM scope ('@leeguoo').
[DYNAMIC_EXECUTION]: The skill provides a self-update command (zentao self-update) that allows the CLI tool to download and execute updates for itself at runtime.
[CREDENTIALS_UNSAFE]: The documentation and capability map suggest passing sensitive ZenTao passwords directly as command-line arguments using the --zentao-password flag. This practice can expose secrets in process monitoring tools, shell history, or logs.
[INDIRECT_PROMPT_INJECTION]: The skill reads untrusted content from the ZenTao platform, such as bug titles, task descriptions, and documentation, which could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: Data is ingested via commands like zentao bugs list, zentao task get, and zentao docs get (SKILL.md).
Boundary markers: The skill does not implement explicit delimiters or boundary markers when presenting external data to the agent context.
Capability inventory: The skill possesses extensive write capabilities, including the ability to resolve bugs, close tasks, and update project metadata.
Sanitization: No sanitization or content validation for the ingested ZenTao data is specified.

zentao