本文へスキップ
Kitbay
スキル

hyperplan

MIT2026年11月16日 更新360GitHubスター

A skill from the OmAgents toolkit for stress-testing a plan or design before it is built.

スキルの内容

A skill from the OmAgents toolkit for stress-testing a plan or design before it is built. The agent queues three critics in a loop tracked by a loop_engine.py script and starts them as parallel background subagents, each given the plan and a checklist: security, covering input validation, auth bypass, data exposure, injection and dependencies; architecture, covering coupling, scale, over-engineering, error handling and testability; and edge cases such as empty input, races, network failures, resource exhaustion and boundary values. Each critic is marked complete or failed, and once all finish the agent merges the findings into one report grouped from critical to low, with locations and fixes. Critical findings block implementation.

できること

  • Run security, architecture and edge-case critics in parallel
  • Track critics in a task loop
  • Merge findings into a report by severity