Reeve
Add persistent memory to any AI agent in one line.
from reeve import ReeveAgent
agent = ReeveAgent(llm=llm)
agent.chat("I prefer Python over JavaScript")
print(agent.chat("What programming language do I prefer?"))
Python
No manual retrieval. No manual storage. No prompt engineering.
Reeve automatically retrieves, updates, and shares memory across AI agents.
Why Reeve?
Most AI applications rely on:
- Chat history
- Vector databases
- Custom memory prompts
- Hand-written retrieval pipelines
As applications grow, these approaches become difficult to maintain and often fail to preserve long-term context.
Reeve provides a persistent memory layer that:
- Stores durable knowledge instead of raw conversation logs
- Automatically retrieves relevant memory before model calls
- Extracts and stores new knowledge after conversations
- Shares memory across agents, frameworks, and models
- Works with existing AI stacks without changing application logic
Quick Start
Install Reeve:
pip install reeve
One-Line Agent Memory
from reeve import ReeveAgent
agent = ReeveAgent(llm=llm)
agent.chat("Our startup uses Memgraph")
print(
agent.chat("What database do we use?")
)
Memgraph
Memory retrieval, storage, and updates happen automatically.
Which API Should I Use?
I want memory in 30 seconds
from reeve import ReeveAgent
agent = ReeveAgent(llm=llm)
I already use LangGraph, LangChain, CrewAI, Agno, or another framework
from reeve import ReeveMemory
agent = Agent(
middleware=[
ReeveMemory()
]
)
I want complete control
from reeve.tools import (
store_memory,
query_memory
)
How Reeve Works
User Message
↓
Retrieve Memory
↓
Inject Context
↓
LLM
↓
Extract Knowledge
↓
Store Memory
Reeve handles the memory lifecycle automatically so developers can focus on building agents instead of memory infrastructure.
Features
-
ReeveAgent for one-line memory integration
-
ReeveMemory middleware for existing agent frameworks
-
Shared memory across agents and models
-
Automatic memory retrieval and injection
-
Durable knowledge extraction
-
Relationship-aware memory graph
-
Entity deduplication and reconciliation
-
Importance-aware memory storage
-
Team and organization memory via namespaces
-
Framework adapters for:
- OpenAI Agents SDK
- LangGraph
- LangChain
- CrewAI
- Agno
-
Hosted MCP support
-
Backward-compatible low-level APIs
Shared Memory Across Agents
Use the same namespace across multiple agents:
from reeve import ReeveMemory
memory = ReeveMemory(
namespace="company_memory"
)
Claude, GPT, Gemini, Cursor, LangGraph agents, and custom applications can share the same memory layer.
Framework Support
OpenAI Agents SDK
from reeve import ReeveMiddleware
from reeve.adapters import OpenAIAgentsReeveAdapter
LangGraph
from reeve import ReeveMemory
LangChain
from reeve import ReeveMemory
CrewAI
from reeve import ReeveMemory
Agno
from reeve import ReeveMemory
Backward Compatibility
Existing integrations continue to work:
from reeve.tools import (
store_memory,
query_memory
)
No migration is required.
Documentation
The full developer guide includes:
- ReeveAgent
- ReeveMemory
- Framework adapters
- Shared memory
- Retrieval policies
- Advanced configuration
- Migration guides
License
MIT
1 maintainer
Metadata
1 maintainer
Release files for reeve 0.1.17
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| reeve-0.1.17.tar.gz | 140.0 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| reeve-0.1.17-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 271.4 kB
Release files / reeve-0.1.17.tar.gz
| Download URL | reeve-0.1.17.tar.gz |
|---|---|
| Size | 140.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
2355f6c60f6d55b6d0e27b2716ef2749c2b3da2159582b10b96c2f1e3fff2184
|
|
BLAKE2b-256 checksum How to use checksums |
b7a31ce950672468564a5b10174dbc2993272e5668c3388d28a12290cfb9c401
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.10.12
|
Release files / reeve-0.1.17-py3-none-any.whl
| Download URL | reeve-0.1.17-py3-none-any.whl |
|---|---|
| Size | 131.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
a621150c3b1b86391cd88a3e291fa58a65b09e8f02f4f38db5c4e81b99ad4a47
|
|
BLAKE2b-256 checksum How to use checksums |
56a114793a1b39044df6f84fb8640e6aff65920f412b5658f6276ccb290ad2c9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.10.12
|
