The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/skill_builder.py utilizes the subprocess.run() function to execute other Python scripts (sync_skills.py and generate_context.py) within the skill's environment to automate the registry synchronization and context update pipeline.
[DYNAMIC_CODE_GENERATION]: The scripts/skill_builder.py script performs dynamic code generation by interpolating user-provided data (such as skill purpose and names) into pre-defined Python templates to create new executable scripts on the filesystem.
[PROMPT_INJECTION]: The skill implements a pipeline where user-provided descriptions of capability gaps are eventually propagated into the project's core configuration files (CLAUDE.md, GEMINI.md, and AGENTS.md) via the sync and context generation scripts. This creates an indirect prompt injection surface where malicious user input could potentially influence the agent's high-level system instructions.
[DATA_EXPOSURE]: The scripts/gap_logger.py script records user requests and conversation contexts into the data/gap-log.json file. This persists conversation history and potential sensitive information in a structured format on the local filesystem.
[PERSISTENCE]: Through scripts/generate_context.py, the skill can modify system-level documentation files that the agent reads at the start of every session. This allows for persistent modification of the agent's behavior and available toolset.

system-awareness