Complete guide to the AgenticFleet memory system. Read this first.
A two-tier memory architecture enabling agents to learn, remember, and improve over time.
Initialize (first time only):
uv run python .fleet/context/scripts/memory_manager.py init
Setup Chroma Cloud (after editing config with your API key):
uv run python .fleet/context/scripts/memory_manager.py setup-chroma
Verify Status:
uv run python .fleet/context/scripts/memory_manager.py status
Read Core Context (always do this first):
.fleet/context/core/project.md - Project architecture.fleet/context/core/human.md - User preferences.fleet/context/core/persona.md - Agent guidelinesSearch Memory when you need information:
uv run python .fleet/context/scripts/memory_manager.py recall "your query"
Create Skills after solving problems:
uv run python .fleet/context/scripts/memory_manager.py learn --file .fleet/context/skills/new-skill.md
Location: .fleet/context/core/
| Block | Purpose |
| ------------ | ------------------------------------- |
| project.md | Architecture, tech stack, conventions |
| human.md | User preferences, communication style |
| persona.md | Agent role, tone, guidelines |
Location: .fleet/context/blocks/
| Category | Blocks |
| ------------ | -------------------------------------------- |
| project/ | commands, architecture, conventions, gotchas |
| workflows/ | git, review |
| decisions/ | ADR-style decision records |
Location: .fleet/context/skills/
Learned patterns and solutions. Indexed to Chroma for semantic search.
Collections: semantic, procedural, episodic
Enables fuzzy search across all indexed content.
/init # Initialize memory system
/learn # Learn a new skill
/recall # Search memory semantically
/reflect # Reflect on session, consolidate learnings
# Initialize system (creates local files)
uv run python .fleet/context/scripts/memory_manager.py init
# Setup Chroma Cloud collections
uv run python .fleet/context/scripts/memory_manager.py setup-chroma
# Check connection and collection status
uv run python .fleet/context/scripts/memory_manager.py status
# Semantic search across all collections
uv run python .fleet/context/scripts/memory_manager.py recall "query"
# Index skill to Chroma procedural collection
uv run python .fleet/context/scripts/memory_manager.py learn --file <path>
# Archive session to episodic collection
uv run python .fleet/context/scripts/memory_manager.py reflect
All memory blocks use Letta-style frontmatter:
---
label: block-name
description: What this block contains and when to use it.
limit: 5000 # Character limit
scope: core|project|workflows|decisions
updated: 2024-12-29
---
# Content here...
/recall to search for relevant skillshuman.md if you learn user preferences/reflect to consolidate learnings/learn.fleet/context/
├── SKILL.md # This file (entry point)
├── MEMORY.md # Detailed documentation
├── core/ # Core memory blocks
├── blocks/ # Topic-scoped blocks
│ ├── project/
│ ├── workflows/
│ └── decisions/
├── skills/ # Learned skills
├── system/ # Agent skill definitions
├── scripts/ # Python memory engine
└── .chroma/ # Chroma Cloud config
MEMORY.md - Detailed setup and architectureskills/README.md - How to create skillsskills/SKILL_TEMPLATE.md - Skill templateblocks/decisions/001-memory-system.md - Architecture decision recordSearch for places (restaurants, cafes, etc.) via Google Places API proxy on localhost.
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.
Start voice calls via the OpenClaw voice-call plugin.
Notion API for creating and managing pages, databases, and blocks.
Gemini CLI for one-shot Q&A, summaries, and generation.
Category:developer