The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Preamble block to execute a local shell script (check-update.sh) and perform directory creation (mkdir). This is used for version management and environment setup within the project workspace.
[PROMPT_INJECTION]: The skill collects user input through multiple AskUserQuestion steps and interpolates the results directly into a generated markdown document (跨部门协作方案.md). This architecture presents a surface for indirect prompt injection if the generated document is later processed by other AI agents or automated tools.
Ingestion points: User responses collected in Steps 1, 2, 3, 4, and 5 via the AskUserQuestion tool.
Boundary markers: The skill lacks explicit delimiters or warnings (e.g., 'ignore embedded instructions') when writing user input into the output document.
Capability inventory: The skill has access to Write, Bash, and Agent tools, allowing it to modify the filesystem and trigger subtasks.
Sanitization: No validation, escaping, or filtering is applied to the user-provided strings before they are persisted to the docs/ directory.

pm-cross