The Agent Skills Directory

[SAFE]: The skill defines a coordination pattern for parallel processing. It includes manual and automated verification steps (such as checking commit SHAs and measurement methods) to ensure sub-agent outputs are consistent. The instructions explicitly warn against including raw data dumps in the final report, which serves as a mitigation against potential indirect prompt injection from worker outputs. All file operations and tool invocations (spawning agents) are within the expected scope of a development orchestration tool.

swarm