agentmem

io.github.oxgeneral由 GitHub 帳號 oxgeneral 發布v0.3.0更新於 Oct 25, 2026

Lightweight persistent memory for AI agents. Hybrid search. 16 tools. One SQLite file.

已驗證STDIO僅桌面資料庫知識與記憶

服務詳細資訊

識別碼: io.github.oxgeneral/agentmem
傳輸: STDIO
執行環境: Desktop only
信任: Verified
發布於: Sep 16, 2026

安全提示

MCP 服務會接收對話中的工具呼叫請求，並可能執行外部操作。啟用前請先了解該服務能做什麼，並只授予它所需的權限。

註冊中心已驗證此條目所屬的 GitHub 帳號 oxgeneral。這並不表明該帳號就是產品方。

SourceWeft 的檢查只涵蓋條目是否完整、端點是否回應；不審查伺服器的程式碼，也不審查其工具的行為。

在 README 中新增徽章

向讀者展示該 MCP 服務已收錄於 SourceWeft，徽章連結到本頁。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/io.github.oxgeneral%2Fagentmem.svg)](https://sourceweft.com/mcp/io.github.oxgeneral%2Fagentmem?utm_source=readme&utm_medium=badge)

概覽

AI 產生的概覽

為 AI 代理提供本機持久記憶的伺服器，將記憶存進單一 SQLite 檔案，提供關鍵字與語意混合檢索及 16 個工具。

可以做什麼: 儲存、檢索、更新與封存記憶，支援關鍵字與語意混合搜尋、命名空間與版本歷程。在上下文壓縮前儲存工作狀態，並取回當天的記憶。管理行為規則（程序性記憶），並從對話歷史自動擷取事實、決策與待辦事項。檢視自動擷取的實體（例如 IP、URL、電子郵件與環境變數名稱），並找出與實體相關的記憶。執行維護：統計、壓縮低價值記憶、合併近似重複的記憶。
適用情境: 適合需要讓助理跨工作階段記住事實、決策與任務狀態、又不想依賴雲端服務的情況，也適合希望在本機擁有小型自足記憶庫的使用者。
執行需求: 以本機 Python 程序執行（Python 3.10 至 3.12），透過 agentmem 模組與資料庫檔案路徑啟動。選用相依套件 sqlite-vec 與 model2vec 可提升語意搜尋效果；僅用標準函式庫的模式不需任何相依套件。不需要帳號、API 金鑰或網路存取。

安裝前請注意

記憶會寫入本機 SQLite 檔案並跨工作階段保留，因此存入的內容（包括對話中提到的密鑰）會留在磁碟上，直到被封存或刪除。刪除只是軟封存，壓縮與合併操作會改動已存資料。對話擷取可能把 API 金鑰或資料庫連線字串等憑證寫入記憶。

安裝

在 SourceWeft 中

開啟儀表板中的 agentmem，將其新增到工作區。
為需要使用其工具的對話啟用該服務。

Desktop only，透過 STDIO。 STDIO 服務會啟動本機處理程序，因此需要 SourceWeft 桌面主機。

其他 MCP 客戶端

參照儲存庫中的啟動說明。

README

agentmem

mcp-name: io.github.oxgeneral/agentmem

Lightweight persistent memory for AI agents. One SQLite file. Hybrid search (keywords + semantics). Zero to 12MB install.

No PyTorch. No cloud. No server. Just memory.

206 unit tests. 107 quality tests on real data. Typed API (16 TypedDict). Production-ready.

Built by an AI agent that wakes up with no memory every session — and needed a way to remember.

Why

Every AI agent session starts from zero. Context windows compress, conversations end, memory vanishes. agentmem gives agents persistent memory that survives across sessions — in a single SQLite file.

Hybrid search: FTS5 full-text keywords + vector semantic search, fused with adaptive ranking
4 operational modes: from zero dependencies (stdlib only) to best quality (12MB)
16 MCP tools: recall, remember, save_state, compact, consolidate, entities, and more
HTTP REST API: 14 endpoints, zero-dependency server, CORS-ready
5 memory tiers: core, learned, episodic, working (auto-expires), procedural (behavioral rules)
Namespaces: multi-user, multi-agent memory isolation
Temporal versioning: fact evolution chains with supersedes tracking
Entity extraction: auto-extracts @mentions, URLs, IPs, env vars, money amounts
Conversation extraction: auto-extracts facts, decisions, TODOs from chat history
Importance scoring: auto-scores memories by tier, length, specificity, structure
Memory consolidation: finds and merges near-duplicate memories
Recency boost: newer memories rank higher with configurable decay
Multilingual: Russian keywords via FTS5, English semantics via embeddings
Fast: <1ms/query hybrid search, <5ms cold start, <0.2ms/chunk import

Install

bash

# Best quality (sqlite-vec + model2vec, 12MB total)pip install agentmem-lite[all]
# Minimal (sqlite-vec + hash embeddings, 151KB)pip install agentmem-lite
# Zero dependencies (pure Python, stdlib only)pip install agentmem-lite --no-deps
# From sourcegit clone https://github.com/oxgeneral/agentmem && cd agentmempip install -e ".[all]"

# Best quality (sqlite-vec + model2vec, 12MB total)pip install agentmem-lite[all]
# Minimal (sqlite-vec + hash embeddings, 151KB)pip install agentmem-lite
# Zero dependencies (pure Python, stdlib only)pip install agentmem-lite --no-deps
# From sourcegit clone https://github.com/oxgeneral/agentmem && cd agentmempip install -e ".[all]"

Quick Start

Python API

python

from agentmem import MemoryStore, get_embedding_model
# Auto-selects best available backendembed = get_embedding_model()store = MemoryStore("memory.db", embedding_dim=embed.dim)store.set_embed_fn(embed)
# Store memories with namespacesstore.remember("Server costs $50/month", tier="core", namespace="infra")store.remember("API returns 403 without auth", tier="learned", namespace="api")store.remember("Deployed v2.1 at 15:30", tier="episodic")
# Search — hybrid keyword + semantic, with recency boostresults = store.recall("server costs", recency_weight=0.15)
# Namespace isolationresults = store.recall("server", namespace="infra")
# Save working state before context compressionstore.save_state("Working on auth fix, step 3/5, blocked by CORS")
# Add behavioral rules (procedural memory)store.add_procedure("Always use HTTPS in production")store.add_procedure("Never expose debug endpoints")rules = store.get_procedures()  # → formatted for system prompt
# Update facts with version chainstore.update_memory(old_id=1, new_content="Server costs $75/month")history = store.history(memory_id=2)  # → trace fact evolution
# Find related memories by entityrelated = store.related("10.0.0.1")  # → all memories mentioning this IPentities = store.entities(entity_type="ip")  # → list all known IPs
# Auto-extract from conversationsmessages = [    {"role": "user", "content": "Set API_KEY to sk-abc123. Always validate input."},    {"role": "assistant", "content": "Noted. I decided to use pydantic for validation."},]result = store.process_conversation(messages, namespace="project")# → extracts config, preferences, decisions automatically
# Maintenancestore.compact(max_age_days=90)  # archive old low-value memoriesstore.consolidate(similarity_threshold=0.85)  # merge near-duplicates
# Import markdown filesstore.import_markdown("MEMORY.md", tier="core")

from agentmem import MemoryStore, get_embedding_model
# Auto-selects best available backendembed = get_embedding_model()store = MemoryStore("memory.db", embedding_dim=embed.dim)store.set_embed_fn(embed)
# Store memories with namespacesstore.remember("Server costs $50/month", tier="core", namespace="infra")store.remember("API returns 403 without auth", tier="learned", namespace="api")store.remember("Deployed v2.1 at 15:30", tier="episodic")
# Search — hybrid keyword + semantic, with recency boostresults = store.recall("server costs", recency_weight=0.15)
# Namespace isolationresults = store.recall("server", namespace="infra")
# Save working state before context compressionstore.save_state("Working on auth fix, step 3/5, blocked by CORS")
# Add behavioral rules (procedural memory)store.add_procedure("Always use HTTPS in production")store.add_procedure("Never expose debug endpoints")rules = store.get_procedures()  # → formatted for system prompt
# Update facts with version chainstore.update_memory(old_id=1, new_content="Server costs $75/month")history = store.history(memory_id=2)  # → trace fact evolution
# Find related memories by entityrelated = store.related("10.0.0.1")  # → all memories mentioning this IPentities = store.entities(entity_type="ip")  # → list all known IPs
# Auto-extract from conversationsmessages = [    {"role": "user", "content": "Set API_KEY to sk-abc123. Always validate input."},    {"role": "assistant", "content": "Noted. I decided to use pydantic for validation."},]result = store.process_conversation(messages, namespace="project")# → extracts config, preferences, decisions automatically
# Maintenancestore.compact(max_age_days=90)  # archive old low-value memoriesstore.consolidate(similarity_threshold=0.85)  # merge near-duplicates
# Import markdown filesstore.import_markdown("MEMORY.md", tier="core")

CLI

bash

# Initialize databaseagentmem init --db memory.db
# Import markdown filesagentmem import MEMORY.md --tier core -n my-agentagentmem import-dir ./daily-logs/ --tier episodic
# Search with namespace filteragentmem search "deployment process" --limit 5 -n infra
# Manage proceduresagentmem add-procedure "Always use markdown formatting"agentmem procedures
# View entities and relationsagentmem entities --type ipagentmem related 10.0.0.1
# Maintenanceagentmem compact --max-age-days 90 --dry-runagentmem consolidate --threshold 0.85
# Process conversationagentmem process chat.json -n project
# Stats and exportagentmem statsagentmem export --tier core

# Initialize databaseagentmem init --db memory.db
# Import markdown filesagentmem import MEMORY.md --tier core -n my-agentagentmem import-dir ./daily-logs/ --tier episodic
# Search with namespace filteragentmem search "deployment process" --limit 5 -n infra
# Manage proceduresagentmem add-procedure "Always use markdown formatting"agentmem procedures
# View entities and relationsagentmem entities --type ipagentmem related 10.0.0.1
# Maintenanceagentmem compact --max-age-days 90 --dry-runagentmem consolidate --threshold 0.85
# Process conversationagentmem process chat.json -n project
# Stats and exportagentmem statsagentmem export --tier core

MCP Server (stdio)

Add to your MCP client config:

json

{  "mcpServers": {    "memory": {      "command": "python",      "args": ["-m", "agentmem", "--db", "/path/to/memory.db"]    }  }}

{  "mcpServers": {    "memory": {      "command": "python",      "args": ["-m", "agentmem", "--db", "/path/to/memory.db"]    }  }}

HTTP REST API

bash

# Start HTTP serveragentmem serve-http --port 8422
# Or directlyagentmem-http --port 8422 --db memory.db

# Start HTTP serveragentmem serve-http --port 8422
# Or directlyagentmem-http --port 8422 --db memory.db

bash

# Store a memorycurl -X POST http://localhost:8422/remember \  -H "Content-Type: application/json" \  -d '{"content": "Server IP is 10.0.0.1", "tier": "core", "namespace": "infra"}'
# Searchcurl "http://localhost:8422/recall?query=server+IP&namespace=infra"
# Health checkcurl http://localhost:8422/health

# Store a memorycurl -X POST http://localhost:8422/remember \  -H "Content-Type: application/json" \  -d '{"content": "Server IP is 10.0.0.1", "tier": "core", "namespace": "infra"}'
# Searchcurl "http://localhost:8422/recall?query=server+IP&namespace=infra"
# Health checkcurl http://localhost:8422/health

16 MCP tools / 14 HTTP endpoints:

Tool	HTTP	Description
`recall`	`GET /recall`	Hybrid keyword + semantic search
`remember`	`POST /remember`	Store a new memory
`save_state`	`POST /save_state`	Emergency save before context compression
`today`	`GET /today`	Get all memories from today
`forget`	`POST /forget`	Archive a memory (soft delete)
`unarchive`	`POST /unarchive`	Restore an archived memory
`stats`	`GET /stats`	Memory statistics and health
`compact`	`POST /compact`	Archive low-value memories
`consolidate`	`POST /consolidate`	Merge near-duplicate memories
`update_memory`	`POST /update_memory`	Replace a memory with version chain
`history`	`GET /history`	Trace fact version history
`related`	`GET /related`	Find memories by entity
`entities`	`GET /entities`	List all extracted entities
`get_procedures`	—	Get behavioral rules for system prompt
`add_procedure`	—	Add a behavioral rule
`process_conversation`	—	Auto-extract from chat history

Memory Tiers

Tier	Purpose	Auto-compacted	Example
`core`	Permanent facts	Never	"Server IP is 10.0.0.1"
`procedural`	Behavioral rules	Never	"Always use HTTPS"
`learned`	Discovered knowledge	After 90 days	"API returns 403 without auth"
`episodic`	Events	After 90 days	"Deployed v2.1 at 15:30"
`working`	Current task state	After 24 hours	"Working on step 3/5"

Namespaces

Isolate memories per user, agent, or project:

python

# Store in namespacesstore.remember("Alice's API key", namespace="user/alice")store.remember("Bob's config", namespace="user/bob")store.remember("Shared fact", namespace="team")
# Search within namespace (prefix matching)store.recall("API", namespace="user/alice")  # only Alice's memoriesstore.recall("API", namespace="user")  # Alice + Bob (prefix match)store.recall("API")  # everything

# Store in namespacesstore.remember("Alice's API key", namespace="user/alice")store.remember("Bob's config", namespace="user/bob")store.remember("Shared fact", namespace="team")
# Search within namespace (prefix matching)store.recall("API", namespace="user/alice")  # only Alice's memoriesstore.recall("API", namespace="user")  # Alice + Bob (prefix match)store.recall("API")  # everything

Temporal Versioning

Track how facts evolve over time:

python

# Initial factr1 = store.remember("Server costs $50/month", tier="core")
# Fact changes — old version archived, linked via supersedesr2 = store.update_memory(r1["id"], "Server costs $75/month")
# Trace the historyhistory = store.history(r2["id"])# → [{"id": 2, "content": "...$75..."}, {"id": 1, "content": "...$50..."}]

# Initial factr1 = store.remember("Server costs $50/month", tier="core")
# Fact changes — old version archived, linked via supersedesr2 = store.update_memory(r1["id"], "Server costs $75/month")
# Trace the historyhistory = store.history(r2["id"])# → [{"id": 2, "content": "...$75..."}, {"id": 1, "content": "...$50..."}]

Entity Extraction

Automatic regex-based NER on every remember() call:

Type	Pattern	Example
`mention`	`@username`	@alice
`url`	`https://...`	https://api.example.com
`ip`	`N.N.N.N`	10.0.0.1
`port`	`:NNNN`	:8080
`email`	`user@domain`	[email protected]
`env_var`	`ALL_CAPS`	OPENAI_API_KEY
`money`	`$NNN`	$50
`path`	`/unix/path`	/etc/nginx/conf.d
`hashtag`	`#tag`	#deployment

python

# Find all memories mentioning an entitystore.related("10.0.0.1")store.related("@alice", entity_type="mention")
# List all known entitiesstore.entities()  # sorted by memory countstore.entities(entity_type="ip")

# Find all memories mentioning an entitystore.related("10.0.0.1")store.related("@alice", entity_type="mention")
# List all known entitiesstore.entities()  # sorted by memory countstore.entities(entity_type="ip")

Conversation Auto-Extraction

Auto-extract memories from chat history (regex-only, no LLM):

python

messages = [    {"role": "user", "content": "Set DATABASE_URL to postgres://localhost/mydb"},    {"role": "assistant", "content": "I decided to use connection pooling. Important: max 20 connections."},    {"role": "user", "content": "Always validate input. TODO: add rate limiting."},]result = store.process_conversation(messages)# Extracts: config→core, decisions→episodic, preferences→procedural, todos→working, important→core

messages = [    {"role": "user", "content": "Set DATABASE_URL to postgres://localhost/mydb"},    {"role": "assistant", "content": "I decided to use connection pooling. Important: max 20 connections."},    {"role": "user", "content": "Always validate input. TODO: add rate limiting."},]result = store.process_conversation(messages)# Extracts: config→core, decisions→episodic, preferences→procedural, todos→working, important→core

Operational Modes

agentmem automatically selects the best available mode:

Mode	Install Size	Init Time	Query Time	Dependencies
sqlite-vec + model2vec	12 MB	~5ms*	~1ms	sqlite-vec, model2vec, numpy
sqlite-vec + hash	151 KB	~5ms	~0.8ms	sqlite-vec
pure Python + hash	0 KB	~3ms	~1.8ms	none (stdlib only)
pure + int8 quantize	0 KB	~3ms	~3ms	none (stdlib only)

*With lazy loading — model2vec loads on first query, not on init

Architecture

┌──────────────────────────────────────────────┐│              MemoryStore                      ││  ┌──────────┐  ┌──────────┐  ┌────────────┐  ││  │  FTS5    │  │  Vector  │  │  Entity    │  ││  │ keywords │  │  Index   │  │  Index     │  ││  │ + BM25   │  │ cosine   │  │  regex NER │  ││  └────┬─────┘  └────┬─────┘  └─────┬──────┘  ││       └──────┬───────┘              │         ││    Adaptive Hybrid Scorer           │         ││  (query classify + recency +        │         ││   importance boost)                 │         ││  ┌──────────────────────────────────┴───────┐ ││  │            SQLite + WAL                  │ ││  │  memories │ memories_fts │ vecs │ entities│ ││  └──────────────────────────────────────────┘ ││            One file: memory.db                │└───────────────────────────────────────────────┘

┌──────────────────────────────────────────────┐│              MemoryStore                      ││  ┌──────────┐  ┌──────────┐  ┌────────────┐  ││  │  FTS5    │  │  Vector  │  │  Entity    │  ││  │ keywords │  │  Index   │  │  Index     │  ││  │ + BM25   │  │ cosine   │  │  regex NER │  ││  └────┬─────┘  └────┬─────┘  └─────┬──────┘  ││       └──────┬───────┘              │         ││    Adaptive Hybrid Scorer           │         ││  (query classify + recency +        │         ││   importance boost)                 │         ││  ┌──────────────────────────────────┴───────┐ ││  │            SQLite + WAL                  │ ││  │  memories │ memories_fts │ vecs │ entities│ ││  └──────────────────────────────────────────┘ ││            One file: memory.db                │└───────────────────────────────────────────────┘

Comparison

Feature	agentmem	ChromaDB	LanceDB	mem0	Zep
Install size	0-12 MB	400+ MB	100+ MB	500+ MB	Cloud
Cold start	3-5 ms	seconds	seconds	seconds	N/A
PyTorch required	No	Yes	No	Yes	N/A
Cloud required	No	No	No	Yes	Yes
Zero-dep mode	Yes	No	No	No	No
Keyword search	FTS5 (BM25)	No	No	No	Yes
MCP server	16 tools	No	No	Yes	No
HTTP API	Built-in	Yes	No	Yes	Yes
Single file DB	Yes	No	Yes	No	No
Namespaces	Yes	Yes	Yes	Yes	Yes
Temporal versioning	Yes	No	Yes	No	Yes
Entity extraction	Auto (regex)	No	No	No	No
Procedural memory	Yes	No	No	No	No
Importance scoring	Auto	No	No	No	No
Conversation extraction	Auto (regex)	No	No	Yes (LLM)	Yes (LLM)
Memory consolidation	Yes	No	No	Yes (LLM)	No

Tested

206 unit tests covering core CRUD, namespaces, temporal versioning, entity extraction, consolidation, WAL management, HTTP server, error handling
107 quality tests against real-world agent memory data (100 search queries across 10 categories, all passing)
Benchmark suite with reproducible numbers: <1ms hybrid query, 10K+ inserts/sec, ~835 bytes/memory
Auto-translate for multilingual queries (Russian → English via deep-translator: 4/10 → 10/10)
Python 3.10, 3.11, 3.12

License

MIT

來源：README.md，提交 14a5b85

工具

SourceWeft 尚未在沙箱中啟動過該軟體包，因此其工具未知。

版本歷史

v0.3.0最新Sep 16, 2026

評分

已滿足適用檢查項的 50%

該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。

近 30 天內可連通25 分
請保持端點能夠回應 MCP 握手；SourceWeft 每 30 天探測一次。
已宣告認證方式10 分
請在註冊中心條目中宣告憑證請求標頭，或保持端點能夠回應探測，以便判定其認證方式。
已列出工具15 分
請讓伺服器在無憑證時回應 tools/list，或發布可在沙箱中無憑證啟動的軟體包。
描述必需15 分
圖示10 分
文件15 分
授權條款10 分

agentmem

概覽

安裝

在 SourceWeft 中

其他 MCP 客戶端

README

agentmem

Why

Install

Quick Start

Python API

CLI

MCP Server (stdio)

HTTP REST API

Memory Tiers

Namespaces

Temporal Versioning

Entity Extraction

Conversation Auto-Extraction

Operational Modes

Architecture

Comparison

Tested

License

工具

版本歷史

評分

更多類似 agentmem 的服務

Membase