The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands using the uv tool, including uv init, uv add, and uv run to manage and verify scripts.
[PRIVILEGE_ESCALATION]: The instructions include using chmod +x on dynamically generated Python files to grant execution permissions for Unix-like environments.
[DYNAMIC_EXECUTION]: The skill generates Python source code based on user input and includes a mandatory verification step where the agent is instructed to run the generated script using uv run --script to confirm its behavior.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface where malicious user requirements could influence the generated script that the agent then executes during the verification phase.
Ingestion points: User requirements for script functionality and dependencies provided during the "形式の確認" (format confirmation) and implementation phases described in SKILL.md.
Boundary markers: None explicitly defined to separate user-provided logic from the script template.
Capability inventory: Shell execution (uv, chmod), file writing, and network operations (via uv downloading dependencies).
Sanitization: The skill suggests using "safe input" during the verification step to avoid side effects.

python-inline-script