Layered Memory MCP Server

Extend AI agent memory beyond token limits with a 4-tier knowledge architecture.

中文 | 日本語 | 한국어

The Problem

AI agents have limited memory — typically 2-4KB of persistent context injected every turn. Once it's full, the agent forgets everything else. You can't store project configurations, user preferences, API conventions, or domain knowledge without constantly fighting the space limit.

The Solution

Layered Memory organizes knowledge into 4 tiers, trading immediacy for capacity:

┌─────────────────────────────────────────────────────┐
│  L0 — Index Layer (2-4KB, injected every turn)      │
│  Pure pointers: "what knowledge exists and where"    │
├─────────────────────────────────────────────────────┤
│  L1 — Knowledge Files (unlimited, loaded on-demand)  │
│  Structured markdown: configs, conventions, facts    │
├─────────────────────────────────────────────────────┤
│  L2 — Skills Layer (loaded when needed)              │
│  Procedures, workflows, tool-specific knowledge      │
├─────────────────────────────────────────────────────┤
│  L3 — Raw Sessions (searched rarely)                 │
│  Full conversation history, searchable by keyword    │
└─────────────────────────────────────────────────────┘

L0 is your table of contents. L1 is your bookshelf. L2 is your cookbook. L3 is your diary.

Features

• Smart Knowledge Injection — Inject facts with auto-dedup and section merging (upsert/append/merge), auto L0 sync
• Hybrid Search (RRF) — Combines keyword, fuzzy, BM25, and semantic search with Reciprocal Rank Fusion for robust ranking. Query-type auto-detection optimizes the semantic/keyword balance
• Agent-Agnostic L0 Access — get_l0_index tool lets any MCP agent retrieve the memory index
• Multi-Agent Namespace — Isolate knowledge per agent via LAYERED_MEMORY_NAMESPACE while sharing common knowledge
• Session Scanning — Extract knowledge candidates from recent agent sessions
• Health Validation — Check L0-L1 consistency, file structure, and knowledge quality, plus a rot and governance summary
• Lifecycle Governance — Flags process state left in durable knowledge (TODO / FIXME / next step / pending / 待办 / 待拍板 …) and suggests a structured action: move_to_todo, rewrite_as_current_rule, or keep_with_reason. Marker lists are configurable
• Cleanup Planning — A read-only planner sorts findings into safe mechanical fixes, items that need a decision, and items that must not be auto-fixed; a separate tool applies only the safe ones, dry-run by default
• Write Safety — Whole-file rewrites keep a backup out of tree under <LAYERED_MEMORY_HOME>/data/backups/knowledge/, so the live knowledge directory holds only live files
• Cleaner Vector Recall — Embeddings use a noise-stripped heading (dates, versions, and issue numbers removed); displayed headings and metadata stay unchanged
• Configurable Catch-all Scans — Maintenance and audit scans find "misc"-style files by glob pattern (misc, notes, inbox, scratch, 杂项, …), so you don't need a file named misc.md
• Space Analysis — Monitor memory usage and get optimization suggestions
• Agent-Neutral — Works with any MCP-compatible agent (Hermes, Claude, Cursor, etc.)
• Zero Dependencies — Core engine uses Python stdlib only; only fastmcp for MCP transport
• Privacy First — All data stored locally, no external API calls

Design Note: The hybrid search ranking is inspired by Semble, a fast code search library for agents. We adapted their RRF (Reciprocal Rank Fusion) approach — with multi-signal boosting and query-aware alpha detection — for markdown knowledge retrieval.

Quick Start

Install

pip install layered-memory-mcp

Hermes Agent

Add to ~/.hermes/config.yaml:

mcp_servers:
  layered-memory:
    command: layered-memory-mcp
    timeout: 30

Then, from an agent session:

1. init_framework() — creates the store and reports whether the MEMORY.md write guard is installed.
2. integrate_agent(action="install_guard") — deploys and enables the guard plugin (Hermes only) and silences the periodic memory-review nudge.
3. Restart Hermes so the plugin loads (gateway: hermes gateway restart).

The guard keeps agent-side writes out of the framework-owned MEMORY.md: the native memory tool with target="memory" and direct write_file / patch edits of memories/MEMORY.md are blocked, and the model is pointed at inject_knowledge (L1 knowledge) or target="user" (the user profile) instead. Set write_guard: auto in ~/.layered-memory/config.yaml to have init_framework() install or repair it on its own.

OpenClaw

Install the MCP server, then register it:

pip install layered-memory-mcp

# Register as an MCP server
openclaw mcp set layered-memory --command layered-memory-mcp

Layered Memory complements OpenClaw's built-in vector-based memory:

• OpenClaw memory: semantic search over session transcripts (heavy, needs embeddings)
• Layered Memory: structured keyword search over curated knowledge files (light, instant)
• Use both: OpenClaw for "what did I say about X?" and Layered Memory for "what's the database connection string?"

Claude Desktop

Add to your Claude Desktop MCP config:

{
  "mcpServers": {
    "layered-memory": {
      "command": "layered-memory-mcp"
    }
  }
}

Cursor / Other MCP Clients

# stdio mode (default)
layered-memory-mcp

# HTTP mode
layered-memory-mcp --transport http --port 8080

# Verbose logging
layered-memory-mcp --verbose

Environment Variables

Variable	Description	Default
LAYERED_MEMORY_HOME	Root directory for memory data	~/.layered-memory/
LAYERED_MEMORY_SESSIONS_DIR	Agent sessions directory (auto-detected)	~/.hermes/sessions/
LAYERED_MEMORY_AUTO_SYNC_L0	Auto-sync L0 index after writes	true
LAYERED_MEMORY_DEDUP_THRESHOLD	Similarity threshold for dedup (0.0-1.0)	0.7
LAYERED_MEMORY_L0_FORMAT	L0 index format: hermes or generic	hermes
LAYERED_MEMORY_NAMESPACE	Agent namespace for multi-agent isolation	shared
LAYERED_MEMORY_PROMOTION_WATCH_PATTERNS	Comma-separated catch-all glob patterns for maintenance scans	built-in English + Chinese list

config.yaml

You can also set most behavior in <LAYERED_MEMORY_HOME>/config.yaml. The precedence is: constructor argument, then environment variable, then config.yaml, then the built-in default. Two of the sections that are relevant to knowledge governance:

governance:
  # Each list EXTENDS the built-in Chinese + English defaults.
  open_markers: ["WIP", "follow-up"]          # open process state → move_to_todo
  resolved_markers: ["shipped"]                # resolved history → rewrite_as_current_rule
  keep_reason_labels: ["Retention note"]       # "<label>: <reason>" keeps a section's markers
  disabled_markers: ["TBD"]                    # turn off defaults that are noise for you

promotion:
  watch_domains: ["misc"]                      # exact catch-all file names (also checked on writes)
  watch_patterns: ["misc", "misc-*", "notes", "inbox", "scratch", "杂项"]  # case-insensitive globs, maintenance/audit scans only

Omit a key to keep the defaults. If you set watch_patterns, your list replaces the built-in pattern list. Set it to [] to turn pattern matching off.

Usage

1. Write Knowledge (Recommended)

The inject_knowledge tool is the primary write path for all agents. It handles deduplication, section targeting, and automatic L0 index sync in a single call.

Agent learns: "Production DB is PostgreSQL 15 on prod-db:5432"
→ inject_knowledge(
    domain="infrastructure",
    section="Database",
    content="PostgreSQL 15 on prod-db:5432, connection pool: 20 max",
    mode="upsert"
  )
← Creates/updates infrastructure.md, auto-syncs L0 index

Write modes:

Mode	Behavior
upsert (default)	Replace if similar content exists, append if new
append	Always append, skip dedup check
merge	Combine unique parts from new + existing

2. Read Knowledge

Agent: "What's the database connection string?"
→ recall_knowledge(keyword="database")
← Returns relevant sections from infrastructure.md

3. Validate Health

→ validate_knowledge()
← Checks L0↔L1 consistency, orphaned files, stale entries, file health,
  and adds rot_summary / rot_health_score and governance_summary
  (state markers found in L1, grouped by lifecycle action)

4. Plan and Apply Cleanup

The cleanup workflow keeps mechanical fixes separate from semantic ones:

1. plan_knowledge_cleanup_tool(max_items=50)
   ← read-only JSON with three buckets:
     safe_auto_fixes      move legacy *.md.bak out of knowledge/, re-sync L0,
                          reindex the vector store, copy open markers into TODOs
     needs_user_decision  rewrite resolved history, merge heading families,
                          shrink oversized files, …
     must_not_auto_fix    markers kept on purpose with a keep-reason line

2. apply_safe_knowledge_cleanup_tool(actions=<items from safe_auto_fixes>)
   ← dry run (default): reports what would change, writes nothing

3. apply_safe_knowledge_cleanup_tool(actions=..., dry_run=False)
   ← applies only the four safe action types, and rejects anything else

The apply tool never deletes or rewrites L1 text. When it turns a marker into a TODO, the source line stays in place. Merges, rewrites, deletions, and cross-domain moves are left to the user or agent, and are made through the normal write tools. For embedding-based checks (cross-file duplicates, split candidates), use reconcile_knowledge_tool.

5. Session Compression (Cron Job)

Set up a daily cron to extract new knowledge from conversations:

1. scan_recent_sessions → get session summaries
2. AI analyzes summaries → identifies stable facts
3. New facts → written via inject_knowledge (auto L0 sync)
4. L0 index → always up to date

6. Legacy CRUD (Also Available)

For direct file manipulation:

Tool	Description
create_knowledge_file	Create a new .md file (auto L0 sync)
update_knowledge_file	Overwrite an existing file (auto L0 sync)
delete_knowledge_file	Delete a file (auto L0 sync)

MCP Tools

Read Tools

Tool	Description
recall_knowledge	Search L1 knowledge files by keyword with relevance scoring
get_knowledge_file	Read a specific knowledge file by name
list_memory_stats	Get space statistics, file sizes, and optimization suggestions
scan_recent_sessions	Scan recent sessions for knowledge extraction candidates
search_sessions_by_keyword	Search session content by keyword
get_l0_index	Retrieve the full L0 index (agent-agnostic)

Write Tools

Tool	Description
inject_knowledge	Primary write path — smart injection with dedup, section targeting, auto L0 sync
create_knowledge_file	Create a new .md file (auto L0 sync)
update_knowledge_file	Overwrite an existing file (auto L0 sync)
delete_knowledge_file	Delete a file (auto L0 sync)

Management Tools

Tool	Description
sync_l0_index	Manually rebuild L0 index from L1 files (supports dry_run)
validate_knowledge	Health check: L0↔L1 consistency, file quality, duplicates, plus rot score and governance (state-marker) summary
manage_l0_entry	Add / remove / replace individual L0 entries

Governance & Cleanup Tools

Tool	Description
audit_rot_tool	Read-only rot audit: health score (0-100), oversized files, garbled headings, stale sections, cross-file duplicates, catch-all promotion candidates
reconcile_knowledge_tool	Embedding-based cross-file duplicate and split-candidate scan
plan_knowledge_cleanup_tool	Read-only cleanup plan: safe_auto_fixes / needs_user_decision / must_not_auto_fix
apply_safe_knowledge_cleanup_tool	Applies only safe mechanical actions from the plan (migrate_legacy_backups, sync_l0_index, reindex_vector_store, create_todo_from_marker); dry_run=True by default

These tools are additive. Existing tools keep their parameters and response fields. validate_knowledge and the post-write governance report only gain extra keys, for example a per-finding action.

MCP Resources

Resource	Description
memory://status	Overall system status and configuration
knowledge://files	List all knowledge files with metadata

MCP Prompts

Prompt	Description
knowledge_compression_prompt	Template for AI-driven knowledge extraction from sessions
cognitive_decision_prompt	Decision framework for disciplined memory usage

Architecture Deep Dive

Why 4 Tiers?

Tier	Cost	Capacity	Use Case
L0 (Index)	Tokens per turn	~2KB	Quick lookup table
L1 (Knowledge)	1 file read	Unlimited	Structured facts
L2 (Skills)	1 skill load	Unlimited	Procedures
L3 (Sessions)	Full search	Unlimited	Historical recall

Write-Once, Fully-Visible Pipeline (v0.5.0)

The key innovation in v0.5.0 is that every write path automatically syncs the L0 index:

Agent calls inject_knowledge(domain="infra", section="Proxy", content="...")
  │
  ├─ 1. Dedup check (SequenceMatcher, threshold=0.7)
  ├─ 2. Resolve action: upsert / append / merge / skip
  ├─ 3. Section targeting (finds or creates ## heading)
  ├─ 4. File write (with fcntl.flock for concurrency safety)
  └─ 5. Auto L0 index sync
        │
        ↓
  L0 index updated → Agent sees it next turn

This eliminates the "write-but-invisible" problem where agents write L1 files but the L0 index (injected into every turn) doesn't update, causing future sessions to ignore the new knowledge.

Durable Knowledge vs. Process State

L1 is for durable knowledge, meaning the rules and facts that hold now. Unfinished work ("TODO: wire up X", "待拍板") and narratives of how something got fixed are process state, and they go stale fast. After a write, and in validate_knowledge, governance scans L1 for state markers and classifies each marker line:

Action	Meaning
move_to_todo	Open work. It belongs in the TODO store (add_todo / list_todos), not in L1
rewrite_as_current_rule	Resolved history ("has been fixed", "已修复"). Rewrite it as the rule that now holds
keep_with_reason	The section has an explicit <keep-reason label>: <reason> line, so its markers are intentional

Detection is deterministic (regex and string matching, with no model calls) and advisory. The marker vocabulary is data and is configured through governance: in config.yaml, so other languages and house styles can be added without code changes.

Derived Data: L0 Index and Vectors

The L1 markdown files are the source of truth. The L0 index and the vector store (data/vectors.db) are derived from them and can be rebuilt at any time (sync_l0_index, reindex_vector_store_tool). The text that gets embedded uses a noise-stripped heading: ## Release process 2026-10-01 (v3.2) is embedded as Release process, so periodic entries cluster by topic rather than by date. Displayed headings and stored metadata keep the original text. After upgrading, you can reindex so that existing vectors pick up the cleaner headings.

Safe Cleanup Boundary

Automatic cleanup is limited to mechanical, reversible, or idempotent actions: moving legacy *.md.bak files out of knowledge/ into data/backups/knowledge/, rebuilding derived L0 and vector data, and copying an open marker into the TODO store (with a deterministic id, so re-applying does nothing). Anything semantic is outside the safe boundary and is always left to the user or agent: deleting facts, merging conflicting sections, moving content across domains, rewriting history as a rule, or removing the source line after a TODO is created. apply_safe_knowledge_cleanup_tool rejects these actions by name.

Relevance Scoring

When you call recall_knowledge, files are scored by:

1. Filename match (+10 points) — keyword appears in filename
2. Heading match (+3 points) — keyword appears in a ## heading
3. Content frequency (+0.5 per occurrence, capped at 5) — how often keyword appears

Results are sorted by score, and only matching ## sections are returned (not entire files).

Namespace Isolation (v0.6.0)

Set LAYERED_MEMORY_NAMESPACE to isolate knowledge per agent:

knowledge/
├── shared/           ← Common knowledge, visible to all agents
│   ├── infrastructure.md
│   └── coding-standards.md
├── claude/           ← Claude Desktop's private knowledge
│   └── claude-specific.md
├── cursor/           ← Cursor IDE's private knowledge
│   └── cursor-config.md
└── hermes/           ← Hermes Agent's private knowledge
    └── hermes-setup.md

Each agent sees its own namespace first, then falls back to shared/. File name collisions resolve in favor of the namespace. All read/search/inject tools automatically merge both directories.

# Claude Desktop config
LAYERED_MEMORY_NAMESPACE=claude layered-memory-mcp

# Backward compatible: default "shared" = no isolation
layered-memory-mcp

L0 Index Formats

Two index formats are supported:

Format	Example	Best For
hermes	[L0索引] infra: servers, DB → knowledge/infra.md	Hermes Agent memory injection
generic	[infra.md] Server Configuration → proxy, db, deploy	Standalone / other agents

Configure via LAYERED_MEMORY_L0_FORMAT env var or the l0_format constructor argument.

Session Compression

The scan_recent_sessions tool is designed for cron-job automation:

1. It scans session files from the past N days
2. Extracts user messages, assistant topics, and tool calls
3. Returns a structured JSON for an AI to analyze
4. The AI identifies stable knowledge and writes it to L1 files via inject_knowledge

This creates a self-improving memory system — the agent gets smarter over time as more knowledge is distilled from conversations.

Agent Compatibility

Layered Memory is an MCP server — it works with any MCP-compatible agent.

Agent	Config Method	Notes
Hermes Agent	config.yaml → mcp_servers	Native MCP client, L0 auto-injection via memory
OpenClaw	openclaw mcp set	Complements built-in vector memory
Claude Desktop	claude_desktop_config.json	Full MCP support, L0 via tool calls
Cursor	Settings → MCP	Full MCP support
Codex CLI	Codex MCP config	Full MCP support
Any MCP client	stdio or HTTP transport	Standard MCP protocol

When to use Layered Memory vs. built-in memory

Most agents have limited persistent memory (2-4KB per turn). Layered Memory solves this by:

1. Separating index from content — L0 stays small (fits in agent memory), L1 holds unlimited knowledge
2. On-demand loading — the agent only reads what it needs, when it needs it
3. Self-improving — session compression automatically extracts new knowledge over time

Integration patterns

Agent (2KB memory limit)
  └── L0 index (injected every turn, ~500 bytes)
        ├── [L0] infrastructure: servers, DB → knowledge/infrastructure.md
        ├── [L0] api: REST conventions → knowledge/api-conventions.md
        └── [L0] dev: code style, testing → knowledge/development.md
              │
              ↓ (on demand via recall_knowledge)
        L1 knowledge files (unlimited, loaded by keyword)

Cognitive Decision Framework

The 4-tier architecture only works if the agent follows a disciplined decision process. This framework should be injected into the agent's system prompt (or loaded via the cognitive_decision_prompt MCP prompt) to ensure consistent behavior.

Decision Tree

Agent encounters a problem or receives a request
  │
  ├─ Step 1: Scan L0 index for relevant domains
  │
  ├─ Step 2: Match found?
  │   ├─ YES → Load the corresponding L1 knowledge file / L2 skill
  │   │   │
  │   │   ├─ Knowledge solves it → Use it. Do NOT bypass with guessing.
  │   │   ├─ Knowledge partially covers it → Use it, then enhance the entry.
  │   │   └─ Knowledge insufficient → Treat as new problem (Step 3).
  │   │
  │   └─ NO → Treat as new problem (Step 3).
  │
  ├─ Step 3: Handle as new problem/requirement
  │   Use standard tools and reasoning to solve.
  │
  └─ Step 4: Post-solution evaluation
      Is this worth preserving?
      ├─ YES → Write to L1 (via inject_knowledge) or L2 (skill) for future reuse.
      └─ NO  → Done.

Why This Matters

Without this decision framework, agents tend to:

• Ignore existing knowledge — they see the L0 index but forget to load L1 files, then waste time guessing
• Repeat mistakes — solved problems aren't captured, so the agent re-learns from scratch next session
• Bypass established conventions — each session starts from zero instead of building on accumulated knowledge

The framework turns the memory system from a passive storage into an active cognitive loop: consult → act → learn → improve.

Integration

Add this to your agent's system prompt:

You use a 4-tier layered memory system. Before tackling any problem:
1. Check L0 index for matching domains
2. If matched, load and follow L1/L2 before acting
3. If unmatched, solve normally
4. After solving, use inject_knowledge to preserve new knowledge

Or use the built-in MCP prompt cognitive_decision_prompt to get the full decision framework at runtime.

Development

# Clone
git clone https://github.com/LAIguapi/layered-memory-mcp.git
cd layered-memory-mcp

# Install in dev mode
pip install -e ".[dev]"

# Run tests
pytest

# Run locally
python -m layered_memory_mcp.server

Changelog

See CHANGELOG.md for the full release history. Recent highlights in 3.4.7:

• Lifecycle governance with structured actions (move_to_todo / rewrite_as_current_rule / keep_with_reason) and configurable marker lists
• plan_knowledge_cleanup_tool and apply_safe_knowledge_cleanup_tool (read-only planning, plus dry-run-first application of safe actions)
• Knowledge backups moved out of the live knowledge directory into data/backups/knowledge/
• Noise-stripped headings in embedded vector text
• Glob-based catch-all scans (promotion.watch_patterns) with English and Chinese defaults

License

MIT License — see LICENSE for details.