The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the dws binary to perform administrative tasks, including searching, retrieving, and installing skills. These operations are executed via subcommands like dws skill search and dws skill install.
[EXTERNAL_DOWNLOADS]: Fetches skill metadata and packages from the official DingTalk API host at mcp.dingtalk.com. This is the core functionality for synchronizing with the DingTalk marketplace and enterprise skill libraries.
[PROMPT_INJECTION]: The skill ingests untrusted data in the form of marketplace search results (skill names and descriptions) which could contain adversarial instructions (Indirect Prompt Injection).
Ingestion points: Data is ingested via the dws skill search command as described in SKILL.md and references/skill.md.
Boundary markers: No explicit delimiters are specified for encapsulating the marketplace search output when displayed to the agent.
Capability inventory: The skill has the capability to write files to local agent configuration directories (e.g., ~/.claude/skills/, ~/.cursor/skills/) during installation.
Sanitization: The skill documentation mentions a securityStatus check (passed, failed, checking) provided by the backend to help the agent and user identify safe content before installation.

dingtalk-skill