team-brainstorm

The skill presents a coherent framework for coordinated multi-role brainstorming with generator-critic loops and shared artifacts. Its footprint is primarily orchestration and internal data handling, which aligns with its stated purpose. However, the ability to autonomously spawn and manage multiple background workers and the breadth of allowed-tools introduce elevated risk if misused or misconfigured. The data flows are largely internal to the skill’s workspace, with no evident external exfiltration in the provided description, but the high degree of autonomy and cross-role interaction warrants cautious governance, access controls, and operational oversight. Overall, the skill is Suspicious due to its high autonomy and potent orchestration capabilities, but not evidently malicious from the described design.

The coordinator specification itself does not contain explicit malicious code. Primary security concerns are delegation and trust: it reads external command files and spawns powerful team-worker agents that can modify filesystem artifacts and run arbitrary workflows. If command files or worker implementations are compromised, an attacker could induce file manipulation, exfiltration, or other harmful actions. Mitigations: validate inputs, protect command/role-spec files (integrity checks and restricted write access), sandbox worker capabilities, and add authorization/audit controls around archive/export/team lifecycle operations.