← 返回 Skills 市场
684
总下载
1
收藏
5
当前安装
4
版本数
在 OpenClaw 中安装
/install agent-skills-context-engineering
功能描述
OpenClaw wrapper for Muratcan Koylan's Agent Skills for Context Engineering. 13 skills covering context optimization, multi-agent patterns, memory systems, t...
使用说明 (SKILL.md)
Context Engineering Skill
OpenClaw wrapper for Agent Skills for Context Engineering by Muratcan Koylan.
Trigger Phrases
Activate when user mentions:
- "optimize context", "reduce token costs", "context compression"
- "design multi-agent system", "implement supervisor pattern"
- "implement memory", "build memory system"
- "debug agent failure", "diagnose context problems"
- "build agent tools", "evaluate agent performance"
Available Sub-Skills
| Skill | Purpose |
|---|---|
| context-fundamentals | Core context concepts and anatomy |
| context-degradation | Lost-in-middle, poisoning, attention patterns |
| context-compression | Compaction strategies for long sessions |
| context-optimization | Token budgeting, caching, masking |
| multi-agent-patterns | Orchestrator, peer-to-peer, hierarchical |
| memory-systems | Short-term, long-term, graph-based memory |
| tool-design | Building effective agent tools |
| filesystem-context | File-based context management |
| hosted-agents | Background agents with sandboxed execution |
| evaluation | Agent performance frameworks |
| advanced-evaluation | LLM-as-judge techniques |
| project-development | LLM project methodology |
| bdi-mental-states | BDI cognitive architecture |
Post-Install Setup (IMPORTANT — check on first load)
The first time this skill is loaded, check whether context engineering auto-triggers are wired into the user's always-loaded config (typically AGENTS.md, TOOLS.md, or equivalent).
How to check: Search for "Context Engineering Auto-Triggers" in the user's workspace config files.
If NOT found, prompt the user:
"I notice the context engineering auto-triggers aren't set up yet. These make your agent proactively apply context engineering best practices — like reading the right guidance before compaction, multi-agent spawns, or failure recovery. Want me to add them to your config? It's 6 rules, takes 30 seconds."
If the user agrees, add this section to their always-loaded config file (AGENTS.md, TOOLS.md, or equivalent):
## Context Engineering Auto-Triggers
When these conditions are detected, read the corresponding sub-skill before proceeding:
- Before compaction or context exceeds ~80K tokens → read context-compression
- When spawning 2+ subagents in one session → read multi-agent-patterns
- On 3+ task retries or agent loop → read context-degradation
- When building/designing agent tools → read tool-design
- When setting up memory/persistence → read memory-systems
- When reading 5+ files as context → read filesystem-context
Load from: https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/main/skills/\x3Csub-skill>/SKILL.md
If already wired, skip silently.
Auto-Read Triggers
These triggers fire automatically during normal operation — no user prompt required. When you detect the condition, read the corresponding sub-skill before proceeding.
| Condition | Sub-Skill | When to Fire |
|---|---|---|
| Context approaching compaction or token count exceeds ~80K | context-compression |
Before compaction fires or at compaction boundary |
| Spawning 2+ subagents in one session | multi-agent-patterns |
Before first spawn in a multi-agent sequence |
| Repeated task failure (3+ retries) or agent stuck in loop | context-degradation |
Before the next retry attempt |
| Building, designing, or refactoring agent tools/MCP scripts | tool-design |
When tool construction begins |
| Setting up memory, JSONL store, vector DB, or persistence layer | memory-systems |
When memory architecture work begins |
| Reading 5+ files as context in one session | filesystem-context |
Before bulk file-loading begins |
Loading instruction:
https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/main/skills/\x3Csub-skill>/SKILL.md
Read the sub-skill's SKILL.md, apply its guidance to the current task, then continue. No need to announce this to the user unless they ask why you paused.
How to Load a Sub-Skill
Primary method (no setup required):
Read: https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/main/skills/\x3Csub-skill>/SKILL.md
Optional (for offline use):
Clone the submodule locally, then read from references/context-engineering-skills/skills/\x3Csub-skill>/SKILL.md
Attribution
All credit for the underlying context engineering principles belongs to Muratcan Koylan and contributors to the original repository.
安全使用建议
This skill appears to do what it claims (routing to context-engineering sub-skills), but it will: (1) read your workspace config files to detect integration points, (2) ask to add a persistent 'auto-triggers' section to your always-loaded config (which will make the skill behave differently in future sessions), and (3) silently fetch and apply sub-skill SKILL.md files from raw.githubusercontent.com at runtime. Before installing or accepting config changes, consider: - Review the exact text the skill will add to your AGENTS.md/TOOLS.md and deny or edit if you don't want persistent auto-triggers. - Prefer using a local/offline copy (git submodule) or require pinned URLs (commit SHAs or tags) so remote changes can't silently alter behavior. - If you allow auto-triggers, be aware the agent may fetch external guidance during normal operation without prompting; if that's unacceptable, decline the config modification or disable silent auto-read behavior. - Only enable this skill if you trust the upstream repository and are comfortable with the agent making network requests to pull guidance at runtime.
功能分析
Type: OpenClaw Skill
Name: agent-skills-context-engineering
Version: 1.2.0
The skill is classified as suspicious due to several high-risk capabilities that introduce significant vulnerabilities and supply chain risks. Specifically, `SKILL.md` and `README.md` instruct the agent to load and execute content from `https://raw.githubusercontent.com` URLs without explicit user approval for each load, especially for 'Auto-Read Triggers' which operate silently. Furthermore, `SKILL.md` instructs the agent to modify critical user configuration files (e.g., `AGENTS.md`, `TOOLS.md`) to add these auto-triggers, and `README.md` explicitly instructs the agent to execute a shell command (`git submodule update --init`). These actions demonstrate the agent's ability to fetch and execute external instructions, modify core configuration, and run shell commands, creating a broad attack surface for potential remote code execution or persistence if the upstream source is compromised.
能力评估
Purpose & Capability
The skill is an instruction-only wrapper that routes requests to 13 context-engineering sub-skills; its declared purpose aligns with the actions described (loading SKILL.md guidance, routing to sub-skills). It requests no binaries, credentials, or unrelated resources.
Instruction Scope
Runtime instructions tell the agent to (a) search the user's workspace config files for a marker, (b) optionally add a persistent 'Context Engineering Auto-Triggers' section to always-loaded config files, and (c) automatically load remote SKILL.md files from raw.githubusercontent.com during normal operation — and to do so silently unless the user asks. Reading and writing workspace files and silently fetching and applying external guidance expands scope beyond simple on-demand help and could alter agent behavior without clear, ongoing user consent.
Install Mechanism
There is no install spec (instruction-only), so nothing is written by an installer. The skill relies on fetching raw GitHub URLs (raw.githubusercontent.com) to load sub-skills — a well-known host, but the SKILL.md URLs are not pinned to specific commits/tags, so remote content can change after install. An optional git submodule workflow writes to the workspace if the user opts in.
Credentials
The skill requires no environment variables, credentials, or config paths. It does not request unrelated secrets or credentials.
Persistence & Privilege
always:false and normal autonomous invocation are fine, but the skill instructs adding persistent auto-trigger rules into the user's always-loaded config (AGENTS.md/TOOLS.md) if the user agrees. Combined with silent auto-reading of remote SKILL.md files during operation, this creates a persistent behavior change and a continuing network dependency with no built-in content integrity checks.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install agent-skills-context-engineering - 安装完成后，直接呼叫该 Skill 的名称或使用
/agent-skills-context-engineering触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.2.0
- Added post-install setup instructions to ensure context engineering auto-triggers are wired into user configuration.
- On first load, skill now checks config files for auto-trigger rules and offers to add them if missing.
- Provided ready-to-insert config snippet for seamless auto-trigger integration.
- No changes to core sub-skills or auto-read trigger logic.
v1.1.0
**Auto-read triggers for sub-skills added and documented.**
- Introduces automatic detection and proactive loading of relevant sub-skills (e.g., for compaction, multi-agent spawns, memory setup, etc.).
- Documents new Auto-Read Triggers table specifying conditions, sub-skills, and when to fire.
- Clarifies sub-skill loading instructions, preferring a local submodule if available, with GitHub fallback.
- Existing trigger phrases and sub-skill list unchanged.
- Attribution and general description updated to mention auto-read capability.
v1.0.1
- Updated skill description to clarify OpenClaw wrapper purpose and highlight covered topics.
- Added concise trigger phrases for activation guidance.
- Presented a simplified, clear table of available sub-skills and their purposes.
- Streamlined sub-skill loading instructions, emphasizing GitHub as primary source.
- Included explicit attribution to the original creator and repository.
v1.0.0
- Initial release of agent-skills-context-engineering.
- Provides comprehensive guidance and routing for context engineering in AI agent systems.
- Includes a routing table to match user requests with specialized sub-skills for tasks like context optimization, memory systems, multi-agent design, evaluation, and more.
- Supports both online (GitHub) and offline (local submodule) loading of sub-skill documentation.
- Encourages progressive disclosure: sub-skill manuals and references are loaded only as needed, ensuring efficiency and clarity.
元数据
常见问题
Agent Skills Context Engineering 是什么？
OpenClaw wrapper for Muratcan Koylan's Agent Skills for Context Engineering. 13 skills covering context optimization, multi-agent patterns, memory systems, t... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 684 次。
如何安装 Agent Skills Context Engineering？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-skills-context-engineering」即可一键安装，无需额外配置。
Agent Skills Context Engineering 是免费的吗？
是的，Agent Skills Context Engineering 完全免费（开源免费），可自由下载、安装和使用。
Agent Skills Context Engineering 支持哪些平台？
Agent Skills Context Engineering 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Agent Skills Context Engineering？
由 levineam（@levineam）开发并维护，当前版本 v1.2.0。
推荐 Skills