菜单
When two agents produce overlapping skills in one session, consolidate by merging unique content into the highest-confidence existing skill.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Pattern for keeping a rolling temperature-trend sample buffer in Hubitat state, computing slope over a configurable window, and classifying rising/falling/steady/unknown.
Implement Tuya Local v3.3 Groovy drivers on Hubitat using rawSocket, AES-128-ECB, queued retries, and defensive frame parsing.
Use Hubitat async HTTP plus a request queue to authenticate with Cognito, cache tokens in state, refresh proactively, and replay a single 401-failed request.
Never let cached-state dedup bypass session validity in cloud-backed Hubitat drivers.
Standard guard pattern for async HTTP response callbacks in Hubitat drivers
Correct formula for converting lat/lng degree differences to miles, with pole clamp for numerical stability
| name | skill-consolidation |
| description | When two agents produce overlapping skills in one session, consolidate by merging unique content into the highest-confidence existing skill. |
| domain | team-process |
| confidence | low |
| source | earned |
When multiple agents independently produce skills on the same topic in a single session, consolidate rather than keeping duplicates:
git rm.Consolidation reduces skill fragmentation, keeps the skill graph navigable, and preserves the confidence lifecycle (low → medium/high only after repeated independent confirmation).