optimize-agent-context

仓库创建 2026年7月26日最近提交 2 天前SkillHot 收录 2 天前

▸ 精选理由

能显著减少冗余提示并提升工具路由与可解释性，适合 Agent 维护者。

▸ 风险提示

会读取或修改本地配置，需先备份并防止凭据泄露。

这个 Skill 做什么

审计并安全优化本地 AI Agent 的上下文、技能与触发与路由配置。

帮你审计并安全优化本地 AI Agent 的上下文、技能、触发器和路由配置，目的是减少过时或冗余的上下文、提升触发精确度并避免重复调用工具。在 agent 开始带太多历史上下文、规则冲突或路由混乱、表现退化时用它来诊断并做可回滚的优化。特别用 ContextLean 做基准测量，能给出证据支持的前后对比和回滚方案。

▸ 展开 SKILL.md 英文原文

Audit, benchmark, and safely optimize local AI Agent context, instructions, Skills, Context Packs, plugins, hooks, and cross-vendor configuration. Use when ChatGPT/Codex, Claude Code, or another CLI Agent starts with excessive or stale context, routes Skills poorly, repeats tool work, has conflicting rules, or needs evidence-backed before/after optimization with rollback. Do not use for ordinary code edits, model/provider tuning, or unsupported speed claims.

Skill 开发管理上下文精简Agent调优跨厂商通用

Stars

Forks

仓库内 Skill

7 日增星

安装 / 使用

给你的 Agent 一句话（通用）

帮我安装这个 skill：https://raw.githubusercontent.com/Offwhite-Del/contextlean/main/plugins/contextlean/skills/optimize-agent-context/SKILL.md

或 curl 直取 SKILL.md

curl -fsSL "https://raw.githubusercontent.com/Offwhite-Del/contextlean/main/plugins/contextlean/skills/optimize-agent-context/SKILL.md"

SKILL.MD 节选查看完整文件 ↗

# Optimize Agent Context

Use ContextLean as the deterministic measurement, evaluation, and rollback layer. Optimize relevance, freshness, trigger precision, and tool behavior; never equate shorter context with better results.

## Workflow

1. Establish the target Agent, project root, current canonical state, quality priority, and representative tasks. Keep account UI instructions, global/project rules, Skills, runtime configuration, Context Packs, and provider-fixed context separate.
2. Run the bundled CLI read-only:

   ```bash
   node <skill-directory>/scripts/contextlean.mjs doctor --root <project> --json
   ```

   Use `audit --scope repo` when personal Agent configuration is out of sco

via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有