context-optimizer SKILL.md skillA

context-optimizer from huangwb8/skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add huangwb8/skills/context-optimizer@git:20260906.d2681a7

Pins git:20260906.d2681a7 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
huangwb8/skills
Path
skills/alpha/awesome-code/agents/context-optimizer/SKILL.md
Versions recorded
4
Latest version
git:20260906.d2681a7, committed Sep 6, 2026
First version recorded
Aug 25, 2026
Size
14,193 bytes
Repository stars
50
License
MIT

上下文优化专家。专注于长对话中的上下文管理、token 效率和性能优化。解决 lost-in-middle、context poisoning 等问题,提升 AI 代理在复杂任务中的表现。.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# Context Optimizer - 上下文优化专家

## 核心理念

**上下文优化** 是长对话性能的关键:

```
┌─────────────────────────────────────────────────────────┐
│  识别问题 → 压缩历史 → 掩码加载 → 缓存重用 → 性能提升  │
└─────────────────────────────────────────────────────────┘
```

**核心问题**:
- ❌ **Lost-in-Middle**:关键信息被中间内容淹没
- ❌ **Context Poisoning**:冲突信息干扰判断
- ❌ **Distraction**:无关信息浪费 token
- ❌ **Context Clash**:多信息源冲突

---

## 何时使用本技能

在以下场景时激活:

- 长对话导致性能下降
- Context window 接近限制
- AI 遗忘之前的信息
- 提到"上下文"、"token 限制"、"效率"

---

## 上下文问题识别

### 问题 1:Lost-in-Middle

**表现**:
- AI 遗忘对话中间的关键信息
- 首尾信息记住,中间信息遗忘
- 需要重复提供相同信息

**检测**:
```python
def detect_lost_in_middle(conversation: list) -> bool:
    """检测是否出现 lost-in-middle 问题"""
    # 1. 检查对话长度
    if len(conversation) < 10:
        return False

    # 2. 检查是否有重复提问
    questions = [msg for msg in conversation if '?' in msg]
    unique_questions = set(questions)
    if len(questions) > len(unique_questions) * 1.5:
        return True  # 存在重复提问

    # 3. 检查中间内容是否被引用
    middle_start = len(conversation) // 3
    middle_end = len(conversation) * 2 // 3
    middle_content = conversation[middle_start:middle_end]

    # 检查后续对话是否引用中间内容
    later_refs = sum(
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add huangwb8/skills/context-optimizer@sha256:8d638616992aff7a

The line at the top of this page pins the label git:20260906.d2681a7, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_fwesku6r5nuuuvea.svg)](https://markdownregistry.com/a/art_fwesku6r5nuuuvea)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 4 versions of this file, between Aug 25, 2026 and Sep 6, 2026, about one every 4 days.

versioncommittedcommitsizeaudit
git:20260906.d2681a7 latest2026-09-06 d2681a7 14,193 BA view · diff
git:20260905.a346b912026-09-05 a346b91 14,189 BA view · diff
git:20260905.7c0ea7c2026-09-05 7c0ea7c 14,150 BA view · diff
git:20260825.4f192d92026-08-25 4f192d9 13,150 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (14193 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

huangwb8/skills · 50 stars · license MIT · pushed 2026-10-17 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_fwesku6r5nuuuvea
GET https://markdownregistry.com/api/v1/resolve?ref=huangwb8/skills/context-optimizer
GET https://markdownregistry.com/api/v1/blob/8d638616992aff7abbe1a447b5fa8b2fe09fc06ce106b23b1fbe4e4406aa9f54

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from huangwb8/skills

AGENTS.md AGENTS.md file
huangwb8/skills · AGENTS.md
git:20260912.90bbdcd · audit A · 50 stars
CLAUDE.md CLAUDE.md file
huangwb8/skills · CLAUDE.md
git:20260219.a2200bc · audit A · 50 stars
auto-draw-plot SKILL.md skill
huangwb8/skills · skills/alpha/auto-draw-plot/SKILL.md · 根据用户描述生成高质量绘图 prompt,并按通用、roadmap、schematic 模式通过 BenszAPI 直接完成 gpt-image-2.5-flare、gpt-image-2.5-sunburst、gpt-image-2 或…
git:20260910.448010a · audit A · 50 stars
auto-test-code SKILL.md skill
huangwb8/skills · skills/alpha/auto-test-code/SKILL.md · 当用户明确要求测试代码、代码审查或代码自检时使用。系统化发现并验证代码问题。⚠️ 不适用:用户只是想实现或优化功能、询问代码问题,或没有明确测试意图。
git:20260906.d2681a7 · audit A · 50 stars
auto-test-project SKILL.md skill
huangwb8/skills · skills/alpha/auto-test-project/SKILL.md · 当用户明确要求“测试项目”、 “运行 auto-test-project”或“进行项目级测试”时使用。对完整项目执行多轮 A 轮批判性测试与 B 轮质量检查,发现、记录、修复并验证问题。⚠️…
git:20260906.d5147b3 · audit A · 50 stars
auto-test-skill SKILL.md skill
huangwb8/skills · skills/alpha/auto-test-skill/SKILL.md · 当用户明确要求测试 Skill、运行 auto-test 或对 Skill 进行批判性测试时使用。系统化发现、记录并验证 Skill 问题。⚠️ 不适用:用户只是想开发或优化 Skill,或没有明确测试意图。
git:20260915.3df7249 · audit A · 50 stars
awesome-code SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/SKILL.md · 当用户明确要求使用 awesome-code、进行多代理协作或需要复杂任务规划时使用。它负责拆解任务、选择专业角色、设置门禁并审查结果;所有 Agent 的创建、隔离、串行或并行执行都必须交给 `parallel-vibe`。⚠️…
git:20261017.6ee9a5b · audit A · 50 stars
backend-specialist SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/agents/backend-specialist/SKILL.md · 后端开发专家。精通 Node.js/Python/Go/Rust 等后端技术栈,专注于 API 设计、数据库优化、认证授权、微服务架构和性能调优。用于后端服务开发、API 设计和系统架构。
git:20260905.a346b91 · audit A · 50 stars
brainstorming SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/agents/brainstorming/SKILL.md · 当用户明确要求"使用 brainstorming"或"使用 awesome-code"时使用。⚠️ 不适用:用户只是想优化/改进某个功能(应直接修改)、只是询问技能问题(应直接回答)、没有明确使用…
git:20260905.a346b91 · audit A · 50 stars
code-reviewer SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/agents/code-reviewer/SKILL.md · 用于审查已完成的工作、重大功能或合并前的变更,核对需求并按严重程度识别风险。未经审查不得合并。
git:20260906.d2681a7 · audit A · 50 stars
devops-specialist SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/agents/devops-specialist/SKILL.md · DevOps 与运维专家。精通 CI/CD、容器化、编排、基础设施即代码、监控告警和自动化部署。用于构建高效、可靠的软件交付流水线和运维系统。
git:20260905.a346b91 · audit A · 50 stars
documentation-specialist SKILL.md skill
huangwb8/skills · skills/alpha/awesome-code/agents/documentation-specialist/SKILL.md · 文档专家。专注于技术文档编写、API 文档生成、README 优化和文档维护。提供清晰的文档结构、规范的格式和用户友好的内容。
git:20260905.a346b91 · audit A · 50 stars

Every file in huangwb8/skills

Other files named context-optimizer

context-optimizer SKILL.md skill
rohitg00/pro-workflow · skills/context-optimizer/SKILL.md · Optimize token usage and context management. Use when sessions feel slow, context is degraded, or you're running out of…
git:20260718.7f7209d · audit A · 2,922 stars
context-optimizer SKILL.md skill
rjmurillo/ai-agents · .claude/skills/context-optimizer/SKILL.md · Analyze skill content for optimal placement (Skill vs Passive Context vs Hybrid), compress markdown to pipe-delimited…
v1.2.0 · audit A · 48 stars

Browse by kind, by grade A, or by owner.