langchain-goodmem 0.1.0

LangChain integration for GoodMem: long-term agent memory with semantic storage and retrieval.

langchain-goodmem

LangChain integration for GoodMem — long-term agent memory with semantic storage and retrieval.

GoodMem is a memory layer for AI agents that handles embedding, vector search, reranking, and LLM-powered answering server-side. This package exposes GoodMem operations as LangChain BaseTools that can be wired into any LangChain agent or chain.

Installation

pip install langchain-goodmem

Requires Python 3.10+.

Tools

Tool	Description
GoodMemListEmbedders	List available embedder models
GoodMemListSpaces	List all spaces in your account
GoodMemGetSpace	Fetch a single space by ID
GoodMemCreateSpace	Create a new space or reuse an existing one
GoodMemUpdateSpace	Update name / public-read / labels on a space
GoodMemDeleteSpace	Delete a space (cascades to its memories)
GoodMemCreateMemory	Store text or files as memories
GoodMemListMemories	Paginate memories within a space
GoodMemRetrieveMemories	Semantic search with optional reranker / LLM summary
GoodMemGetMemory	Fetch a specific memory by ID
GoodMemDeleteMemory	Permanently delete a memory

Quick start

from langchain_goodmem import (
    GoodMemCreateSpace,
    GoodMemCreateMemory,
    GoodMemRetrieveMemories,
)

goodmem_kwargs = {
    "goodmem_base_url": "http://localhost:8080",
    "goodmem_api_key": "your-api-key",
}

tools = [
    GoodMemCreateSpace(**goodmem_kwargs),
    GoodMemCreateMemory(**goodmem_kwargs),
    GoodMemRetrieveMemories(**goodmem_kwargs),
]

Use with a LangChain agent

from langchain.chat_models import init_chat_model
from langchain.agents import create_agent

llm = init_chat_model("openai:gpt-4o")
agent = create_agent(llm, tools)

response = agent.invoke({"messages": [{"role": "user", "content": "Save this fact and recall it later."}]})

Retrieve with reranking and LLM summary

GoodMemRetrieveMemories exposes the full server-side post-processor. Set any of these fields to enable reranking, threshold filtering, chronological re-sort, or an LLM-generated abstractReply summary:

result = retrieve.invoke({
    "query": "Which framework develops applications with language models?",
    "space_ids": "space-uuid-1,space-uuid-2",
    "max_results": 5,
    "reranker_id": "reranker-uuid",
    "llm_id": "llm-uuid",
    "llm_temperature": 0.2,
    "relevance_threshold": 0.1,
    "chronological_resort": False,
})

Leave them all unset for plain semantic search.

Environment variables

Variable	Description
GOODMEM_BASE_URL	Base URL of the GoodMem API server
GOODMEM_API_KEY	API key for authentication
GOODMEM_VERIFY_SSL	Set to false to skip TLS verification for self-signed dev certs (default: true)

Examples

A live end-to-end smoke test exercising every tool and every post-processor knob is in examples/live_smoke_test.py.

export GOODMEM_BASE_URL=https://localhost:8080
export GOODMEM_API_KEY=...
export GOODMEM_VERIFY_SSL=false
python examples/live_smoke_test.py

License

MIT — see LICENSE.