system-init
Initialize or hydrate the agent's memory system and verify configuration.
When & Why to Use This Skill
The system-init skill is a foundational utility designed to initialize and hydrate a Claude agent's memory system. It automates the setup of directory structures, configuration files, and core context templates, ensuring the agent has a robust framework for semantic, procedural, and episodic memory. By integrating with Chroma Cloud, it enables persistent storage and retrieval of project facts, learned patterns, and session history, significantly enhancing the agent's long-term reasoning and context awareness.
Use Cases
- New Project Onboarding: Automatically generate the necessary file structure and memory templates when starting a new agentic development project to ensure consistency.
- Memory System Repair: Quickly restore or reset the agent's core context files and directory hierarchy if they become corrupted or are accidentally deleted.
- Vector Database Configuration: Streamline the connection to Chroma Cloud by managing API keys, tenants, and collection creation for persistent knowledge storage.
- Session Context Initialization: Set up a dedicated recall scratchpad to track and maintain state across complex, multi-step agent interactions.
| name | system-init |
|---|---|
| description | Initialize or hydrate the agent's memory system and verify configuration. |
System Initialization
This skill handles the initialization of the agent's memory system. It ensures that the necessary directory structures, configuration files, and core context templates are in place.
Usage
Run this command at the start of a new project or when you need to reset/repair the memory system structure.
uv run python .fleet/context/scripts/memory_manager.py init
Actions Performed
- Hydrates Templates: Copies
core/*.template.mdto activecore/*.mdfiles if they don't exist. - Config Setup: Ensures
.chroma/config.yamlexists (creating from template if needed). - Recall Scratchpad: Creates
recall/current.mdfor the current session context.
Chroma Cloud Setup
After initialization, set up Chroma Cloud:
Edit config: Add your Chroma Cloud credentials to
.fleet/context/.chroma/config.yaml:cloud: tenant: "your-tenant" database: "your-database" api_key: "your-api-key"Create collections: Run the setup command to create collections in Chroma Cloud:
uv run python .fleet/context/scripts/memory_manager.py setup-chromaVerify status: Check that everything is connected:
uv run python .fleet/context/scripts/memory_manager.py status
Collections Created
| Collection | Name | Purpose |
|---|---|---|
semantic |
agentic-fleet-semantic | Facts about project and user |
procedural |
agentic-fleet-procedural | Learned skills and patterns |
episodic |
agentic-fleet-episodic | Session history and context |
Environment Variables (Optional)
Instead of config file, you can use environment variables:
CHROMA_API_KEY- Your Chroma Cloud API keyCHROMA_TENANT- Your tenant nameCHROMA_DATABASE- Your database name