Agent Skills for Context Engineering

A comprehensive, open collection of Agent Skills focused on context engineering and harness engineering principles for building production-grade AI agent systems. These skills teach the art and science of curating context, designing agent operating loops, and evaluating agent behavior across any agent platform.

课时

• Skills Overview
  • Foundational Skills
    • context-fundamentals
    • context-degradation
    • context-compression
  • Architectural Skills
    • multi-agent-patterns
    • long-horizon-prompting
    • memory-systems
    • tool-design
    • filesystem-context
    • hosted-agents
  • Operational Skills
    • context-optimization
    • latent-briefing
    • evaluation
    • advanced-evaluation
    • harness-engineering
    • self-improvement-loops
  • Development Methodology
    • project-development
  • Cognitive Architecture Skills
    • bdi-mental-states
• Examples
  • examples
  • digital-brain-skill
  • llm-as-judge-skills
  • book-sft-pipeline
  • interleaved-thinking
  • long-horizon-prompt-lab
  • Digital Brain Skill Example
    • HOW-SKILLS-BUILT-THIS.md
• Researcher Operating System
  • researcher
  • Measured router-benchmark results
    • researcher/benchmarks/router/results-published/2026-05-15.md
    • researcher/benchmarks/router/results-published/2026-05-15-v2.md
    • researcher/benchmarks/router/results-published/2026-05-19.md
  • What it includes
    • live corpus inventory
  • Operator commands
    • researcher/runbooks/continuous-operation.md
• Structure
  • template
• Agent Skills for Context Engineering
• 文档
  • PDF Processing
  • Agent Skills for Context Engineering
  • Agent Skills for Context Engineering
  • Agent Skills for Context Engineering
  • Agent Skills for Context Engineering
  • Agent Skills for Context Engineering
  • Agent Skills for Context Engineering
  • Skills Improvement Analysis: Lessons from Anthropic's "Building Claude Code" Article
  • Agent Skills for Context Engineering
  • Architecture decision records
    • ADR-0001: Machine-testable, deny-by-default authority
    • ADR-0002: Treat the corpus inventory as a checked derived view
    • ADR-0003: Publish allowlisted projections, not redacted private records
    • ADR-0004: Make schemas, identity, and frozen bytes runtime-neutral
  • specs
    • SPEC-000: Program constitution and authority model
    • SPEC-001: Repository reconciliation and generated corpus inventory
    • SPEC-002: Public and private boundary
    • SPEC-003: Schema registry and artifact identity
• 示例
  • book-sft-pipeline
    • Book SFT Pipeline
    • 示例
      • Example: Gertrude Stein Style Transfer
        • Sample Model Outputs
    • references
      • Segmentation Strategies
      • Tinker Format Specification
  • digital-brain-skill
    • Digital Brain - Claude Instructions
    • Digital Brain
    • Skills Mapping: Digital Brain
    • content
      • Content Calendar
      • Content Hub
      • templates
        • LinkedIn Post Template
        • Newsletter Template
        • Thread Template
    • 示例
      • Example: Content Creation Workflow
      • Example: Meeting Preparation Workflow
    • identity
      • Bio Variants
      • Personal Brand Strategy
      • Identity Module
      • Voice & Tone Guide
    • knowledge
      • Competitive Landscape
      • Knowledge Base
      • research
        • Research: [TOPIC]
    • network
      • Introductions Tracker
      • Network Module
    • operations
      • Operations Module
      • Task List
      • reviews
        • Weekly Review: Week of [DATE]
    • references
      • File Format Reference
  • interleaved-thinking
    • Reasoning Trace Optimizer
    • 文档
      • Aligning to What? Rethinking Agent Generalization in MiniMax M2
      • M2.1 Tool Use & Interleaved Thinking
      • Compatible Anthropic API
    • generated_skills
      • comprehensive-research-agent
        • Comprehensive Research Agent Best Practices
  • llm-as-judge-skills
    • Agents Index
      • evaluator-agent
        • Evaluator Agent
      • orchestrator-agent
        • Orchestrator Agent
      • research-agent
        • Research Agent
    • Prompts Index
      • agent-system
        • Orchestrator System Prompt
      • evaluation
        • Direct Scoring Prompt
        • Pairwise Comparison Prompt
      • research
        • Research Synthesis Prompt
    • Skills Index
      • context-fundamentals
        • Context Fundamentals Skill
      • llm-evaluator
        • LLM-Evaluator Skill
      • tool-design
        • Agent Tool Design Skill
    • Tools Index
      • evaluation
        • Direct Score Tool
        • Generate Rubric Tool
        • Pairwise Compare Tool
      • orchestration
        • Delegate to Agent Tool
      • research
        • Read URL Tool
        • Web Search Tool
  • long-horizon-prompt-lab
    • report
      • CDC Report Provenance
  • x-to-book-system
    • PRD: X-to-Book Multi-Agent System
    • Skills Mapping: X-to-Book System
• Governance policy
  • generated
    • Authority table
• researcher
  • The Infinite Software Crisis – Jake Nations, Netflix
  • Agent Skills for Context Engineering
  • Source Registry
  • Private artifact storage
  • Researcher Benchmarks
    • Benchmark Architecture Plan
    • Effectiveness Benchmark (Stage 3)
      • tasks
        • 001 - Filesystem context offload
          • Agent Skills for Context Engineering
          • starting
            • Workspace layout
    • Router Benchmark (Stage 2)
      • Available skills ({{SKILLCOUNT}})
      • Published Router Benchmark Results
    • Researcher SDK Runner
  • Claim Provenance
  • corpus
    • SPEC-001 reconciliation report
  • Public export boundary
    • 示例
      • Restricted citation projection example
  • fixtures
    • Export boundary fixtures
    • skill-proposals
      • Skill Proposal: Harness Engineering From Autoresearch
  • insights
    • Auto-Research Experiment: Lessons From Building The Researcher OS
    • How We Built v2.3.0
  • Mechanism Registry
  • Researcher Queue
  • rubrics
    • Content Curation Rubric
    • Harness Change Rubric
    • Pairwise Skill Revision Rubric
    • Skill Change Rubric
  • runbooks
    • Autonomous Research Loop
    • Constitution amendment and emergency disable
    • PR Readiness Runbook
    • Public export correction and removal
    • Schema migration, quarantine, and rollback
  • Researcher Runs
    • 20260515-035228-executable-autonomous-research-frameworks
      • Research Thread: 20260515-035228-executable-autonomous-research-frameworks
      • proposals
        • Skill Proposal: Executable autonomous research frameworks
      • sources
        • evidence
          • Deep Research Evidence Summary
  • Schema registry and canonical artifacts
    • generated
      • Generated schema compatibility matrix
  • templates
    • Research Thread: [Run ID]
    • Skill Proposal: [Short Title]
• skills
  • advanced-evaluation
    • references
      • Bias Mitigation Techniques for LLM Evaluation
      • Evaluation Pipeline Diagram
      • LLM-as-Judge Implementation Patterns
      • Metric Selection Guide for LLM Evaluation
  • bdi-mental-states
    • references
      • BDI Ontology Core Patterns
      • BDI Framework Integration Patterns
      • BDI RDF Examples
      • SPARQL Competency Queries
  • context-compression
    • references
      • Context Compression Evaluation Framework
  • context-degradation
    • references
      • Context Degradation Patterns: Technical Reference
  • context-fundamentals
    • references
      • Context Components: Technical Reference
  • context-optimization
    • references
      • Context Optimization Reference
  • evaluation
    • references
      • Evaluation Reference: Metrics and Implementation
  • filesystem-context
    • references
      • Filesystem Context Implementation Patterns
  • hosted-agents
    • references
      • Infrastructure Patterns for Hosted Agents
  • latent-briefing
    • references
      • Attention Matching (AM) and Task-Guided Scoring
  • long-horizon-prompting
    • references
      • The Cycle Double Cover Prompt, Annotated
      • Research Evidence for Long-Horizon Brief Elements
      • Pseudo-Formal Task Brief Template
      • Vendor Long-Horizon Prompting Doctrine
  • memory-systems
    • references
      • Memory Systems: Technical Reference
  • multi-agent-patterns
    • references
      • Multi-Agent Patterns: Technical Reference
  • project-development
    • references
      • Case Studies: LLM Project Development
      • Pipeline Patterns for LLM Projects
  • self-improvement-loops
    • references
      • Loop Design Evidence
  • tool-design
    • references
      • Architectural Reduction: Production Evidence
      • Tool Design Best Practices
• template
  • references
    • Reference File
    • Topic Details

开始学习 → PDF Processing