{
  "id": "page:docs-reference-repos-muratcankoylan-agent-skills-for-context-engineering-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/muratcankoylan/agent-skills-for-context-engineering/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/muratcankoylan/Agent-Skills-for-Context-Engineering/research.md",
    "sourceKind": "repo-docs",
    "title": "muratcankoylan/Agent-Skills-for-Context-Engineering",
    "displayName": "muratcankoylan/Agent-Skills-for-Context-Engineering",
    "slug": "docs/reference-repos/muratcankoylan/agent-skills-for-context-engineering/research",
    "articlePath": "wiki/docs/reference-repos/muratcankoylan/Agent-Skills-for-Context-Engineering/research.md",
    "article": "\n# muratcankoylan/Agent-Skills-for-Context-Engineering\n\n- **Archetype**: mega-skill-pack\n- **Stars**: 14,978\n- **Last pushed**: 2026-04-12\n- **License**: MIT\n- **Discovered**: 2026-04-12\n- **Source**: gh-search (keyword: \"agent skill\")\n- **Skills found**: 19+\n\n## Summary\nComprehensive collection of agent skills for context engineering, multi-agent architectures, and production agent systems. Organized into skills (context-fundamentals, context-compression, context-degradation, context-optimization, evaluation, advanced-evaluation, tool-design, hosted-agents, memory-systems, bdi-mental-states, filesystem-context, multi-agent-patterns, project-development) plus examples (book-sft-pipeline, digital-brain-skill, interleaved-thinking). Also includes a skill template.\n\n## Assessment\nHIGH VALUE. This is a methodology-rich collection covering agent system design patterns that overlap significantly with babysitter's architectural concerns. Key extractable content:\n- Context degradation patterns (lost-in-middle, U-shaped attention, context poisoning)\n- Multi-agent coordination patterns (supervisor, swarm, hierarchical)\n- Memory system design (vector RAG, knowledge graphs, filesystem-as-memory)\n- Context compression strategies (tokens-per-task optimization)\n- BDI mental states for agents (belief-desire-intention architecture)\n- Evaluation frameworks for agent systems\n- The book-SFT-pipeline example has a staged idempotent architecture pattern\n\n## Extraction Priority\nHIGH -- Multiple skills map directly to babysitter process methodology areas:\n- context-compression -> methodologies/ (complements babysitter's compression layer)\n- multi-agent-patterns -> methodologies/ (orchestration patterns)\n- evaluation + advanced-evaluation -> methodologies/ (quality assessment)\n- memory-systems -> specializations/shared/ (cross-domain pattern)\n- project-development -> methodologies/ (task-model fit analysis)\n\n## Processes\n1. **context-optimization-workflow** -- Systematic context window optimization: compaction, observation masking, prefix caching, strategic partitioning\n2. **agent-evaluation-framework** -- Multi-dimensional agent evaluation: factual accuracy, completeness, tool efficiency, process quality\n3. **multi-agent-architecture-design** -- Pattern selection for multi-agent systems (supervisor vs swarm vs hierarchical)\n4. **memory-system-design** -- Choose and implement agent memory architecture (scratchpad -> temporal knowledge graph spectrum)\n5. **book-sft-pipeline** -- Staged idempotent pipeline: acquire -> prepare -> process -> parse -> render\n6. **digital-brain-skill** -- Personal knowledge management with agents (contacts, content, weekly reviews)\n\n## Plugin Ideas\n- **context-engineering-advisor plugin**: Babysitter plugin that analyzes process definitions for context engineering anti-patterns and suggests optimizations\n\n## Implicit Procedural Knowledge\n- Filesystem-as-memory pattern: using ls/glob/grep/read_file for targeted context discovery outperforms semantic search for structural queries\n- Context isolation as the primary reason for sub-agents (not role simulation)\n- Token-per-task as the correct optimization target (not tokens-per-request)\n- Warm sandbox pools and predictive warming for hosted agent infrastructure\n- Tool consolidation principle: prefer single comprehensive tools over multiple narrow ones\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| Context Optimization Workflow | UPGRADE | Enhanced context window optimization strategies | library/specializations/shared/context-compression.js | specializations/shared/context-compression.js |\n| Agent Evaluation Framework | NEW | Multi-dimensional agent evaluation methodology | - | methodologies/agent-evaluation/ |\n| Multi-Agent Architecture Design | UPGRADE | Pattern selection for multi-agent coordination | library/specializations/shared/orchestration-patterns.js | specializations/shared/orchestration-patterns.js |\n| Memory System Design | NEW | Agent memory architecture selection and implementation | - | specializations/shared/memory-system-design.js |\n| Context Degradation Prevention | NEW | Lost-in-middle and context poisoning mitigation strategies | - | specializations/shared/context-degradation-prevention.js |\n| BDI Mental States Architecture | NEW | Belief-desire-intention architecture for agent systems | - | specializations/ai-agents-conversational/bdi-mental-states.js |\n| Filesystem-as-Memory Pattern | NEW | Structural query optimization using filesystem operations | - | specializations/shared/filesystem-as-memory.js |\n| Tool Design Optimization | NEW | Tool consolidation and efficiency principles | - | specializations/shared/tool-design-optimization.js |\n| Hosted Agents Infrastructure | NEW | Warm sandbox pools and predictive warming patterns | - | specializations/devops-sre-platform/hosted-agents-infrastructure.js |\n| Project Development Task-Model Fit | NEW | Task-model fit analysis for project development | - | specializations/shared/project-development-task-model-fit.js |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| Context Engineering Advisor | NEW | Analyze process definitions for context anti-patterns and optimization | - | plugins/a5c/marketplace/blueprints/context-engineering-advisor/ |\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-muratcankoylan-agent-skills-for-context-engineering-research",
      "kind": "contains_page"
    }
  ]
}