construct

No explicit covert malware behavior (exfiltration/persistence/obfuscation) is evident in this module. However, it provides high-impact arbitrary shell command execution: it runs commands extracted from Blueprint content using subprocess.run(..., shell=True) with only minimal string validation. The workspace snapshot/authorization mechanisms reduce unintended file mutations but do not mitigate the fundamental RCE capability. Treat this as a substantial supply-chain security risk when Blueprint/bundle inputs cannot be fully trusted and sandboxed.

No direct signs of embedded malware (no network calls, no credential theft/exfiltration, no obfuscated payload decoding, and no eval/exec/persistence in this module). However, this is a high-impact workflow adapter that validates a manifest and then executes subprocess argv defined by the installed skill and its manifest—so the main security risk is supply-chain arbitrary code execution if the installed skill/manifest is malicious. Path/template/state integrity checks provide guardrails but do not sandbox or restrict the executed programs beyond limited path/script checks.