The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script manage_proposals.py dynamically loads a module from a path computed at runtime (../../guide-planning/scripts/scope_runtime.py). While this is used for code sharing within the vendor's skills, dynamic loading from relative paths is a potential vector for executing unexpected code.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input for summaries and review notes which are saved to Markdown and JSON files. \n
Ingestion points: manage_proposals.py command-line arguments. \n
Boundary markers: Absent in generated files. \n
Capability inventory: File system write access via management scripts. \n
Sanitization: Input is not filtered for prompt injection patterns, potentially affecting downstream processes that read these files.
[COMMAND_EXECUTION]: The skill relies on the execution of local Python scripts to manage file operations and registry state. While these scripts are bundled with the skill, they provide a mechanism for the agent to modify the repository file system based on user-influenced parameters.

propose