Filesystem Context
Skill Verified ActiveThis skill should be used when the user asks to "offload context to files", "implement dynamic context discovery", "use filesystem for agent memory", "reduce context window bloat", or mentions file-based context management, tool output persistence, agent scratch pads, or just-in-time context loading.
To enable agents to manage and utilize context effectively by leveraging the filesystem for persistent storage, dynamic loading, and efficient information retrieval, thereby overcoming context window limitations.
Features
- Offload large tool outputs to files
- Persist agent plans and state
- Manage agent workspaces via filesystem
- Dynamically load skills from files
- Combine filesystem search with semantic search
Use Cases
- Reducing context window bloat by offloading large outputs
- Implementing agent scratch pads for intermediate results
- Enabling sub-agents to share information via file workspaces
- Persisting agent plans across long-horizon tasks
Non-Goals
- Replacing prompt engineering entirely
- Handling context that fits comfortably within the window
- Adding measurable latency for critical tasks
- Managing context for agents without filesystem tool capabilities
Workflow
- Determine if tool output exceeds token threshold
- Offload large outputs to a designated scratch directory
- Return a compact reference (path, summary) to context
- Persist agent plans and state to structured files
- Use agent workspaces for inter-agent communication
- Dynamically load skill definitions from files as needed
Practices
- Context Management
- Agent Memory
- Persistence
- Tool Design
Prerequisites
- Python 3.7+
- Basic filesystem access
Installation
First, add the marketplace
/plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering/plugin install Agent-Skills-for-Context-Engineering@context-engineering-marketplaceQuality Score
VerifiedTrust Signals
Similar Extensions
Claude Memory Kit
98Persistent memory for Claude Code agents with an agent-audit-ritual architecture. User only talks; agent captures, audits, proposes promotions, and writes. Four layers — daily logs, hot cache (MEMORY.md), role-based reference skills (Anthropic-native, user-invocable: false), canonical rules. /close-day runs the audit ritual at end of day. Multi-project isolation via projects/<name>/. Zero external dependencies.
Context Mode
100Update context-mode from GitHub and fix hooks/settings. Pulls latest, builds, installs, updates npm global, configures hooks. Trigger: /context-mode:ctx-upgrade
Ctx Doctor
100Run context-mode diagnostics. Checks runtimes, hooks, FTS5, plugin registration, npm and marketplace versions. Trigger: /context-mode:ctx-doctor
Memory Management
100Persist SEO/GEO campaign context across Claude sessions with automatic hot-list, active work, and archive tiers. 项目记忆/跨会话
Horizon Track
100Track long-horizon objectives across multiple sessions with milestone checkpoints, progress persistence, and drift detection
Embedding Strategies
100Select and optimize embedding models for semantic search and RAG applications. Use when choosing embedding models, implementing chunking strategies, or optimizing embedding quality for specific domains.