The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the vendor-provided postplus CLI and a local Python script (scripts/workspace.py) to manage project configuration. These tools operate on local files and do not require elevated privileges or access to sensitive system paths.
[SAFE]: Indirect Prompt Injection Assessment: (1) Ingestion points: Untrusted project files such as product.md and project.yaml are read by the workspace script. (2) Boundary markers: The script identifies content using explicit Markdown and YAML keys. (3) Capability inventory: Actions are limited to file creation and readiness validation within the project folder. (4) Sanitization: Custom regex-based parsing ensures that ingested data is treated as static values and not executed as commands.
[SAFE]: The script explicitly avoids system-level paths (home, root) and does not use external dependencies or obfuscation, aligning with secure development practices for local agent skills.

postplus-workspace