The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/_skill_cli.py includes a utility function run_command that wraps subprocess.run for executing system commands.
Evidence: The function uses subprocess.run(cmd, cwd=cwd, capture_output=True, text=True, timeout=timeout) with a list of strings for the command, which is a standard but powerful capability for development-oriented skills.
[INDIRECT_PROMPT_INJECTION]: The skill instructions in SKILL.md direct the agent to execute a local Python script and include a specific portion of its output in the final response to the user.
Ingestion points: The agent is instructed to run python .agents/skills/heyeddi/scripts/suggest_next_skill.py --current-skill motion-physics --project-root . and consume the ### Next step block from the stdout.
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard any potentially malicious directives embedded within the script's output.
Capability inventory: The skill has the capability to execute shell commands via the subprocess.run wrapper in scripts/_skill_cli.py.
Sanitization: The skill does not define any sanitization or validation protocols for the external script output before it is interpolated into the agent's context.

motion-physics