What problem does it solve? Running shell commands from an AI agent is risky without guardrails: commands can hang indefinitely, destructive operations can wipe filesystems, and unstructured output makes failures hard to diagnose. This Skill provides a controlled subprocess execution layer that validates every command against a blocklist, enforces timeouts, and returns structured results with exit codes, stdout, stderr, and timing. ## Core Features & Use Cases - Safe Synchronous Execution: Run commands with a default 60-second timeout, separate stdout/stderr capture, and exit code reporting. - Background Process Management: Start long-running servers or watchers with PID tracking, then terminate them gracefully via SIGTERM with SIGKILL fallback. - Command Blocklisting: Pre-validate commands against dangerous patterns (rm -rf /, fork bombs, firewall flushes, disk formatting) before execution. - Use Case: An orchestrator agent needs to run a test suite, capture failures, and start a local dev server in the background. The Skill runs the tests with a timeout, reports the non-zero exit code with stderr details, starts the server with a tracked PID, and kills it cleanly when the task completes. ## Quick Start Ask the agent to run a shell command such as a build or test script with a timeout and return the exit code, stdout, and stderr as a structured result.
