Skip to main content

​
Install

pip install plyra-memory

​
Remember and recall

import asyncio
from plyra_memory import Memory

async def main():
    async with Memory(agent_id="my-agent") as mem:
        # Write to memory
        await mem.remember("User prefers async Python frameworks")
        await mem.remember("The bug is in the conditional edge logic")

        # Recall — fused search across all layers
        result = await mem.recall("Python preferences")
        for r in result.results:
            print(f"[{r.layer.value}] {r.content}  score={r.score:.2f}")

        # Context — token-budgeted string ready to inject into a prompt
        ctx = await mem.context_for("what stack does the user prefer?")
        print(ctx.content)
        # → "[episodic] User prefers async Python frameworks (score: 0.94)"

asyncio.run(main())

​
Persistent across sessions

The same agent_id retrieves memory from all previous sessions: 
# Session 1
async with Memory(agent_id="support-agent") as mem:
    await mem.remember("User is debugging a LangGraph state machine")

# Session 2 — days later, new process
async with Memory(agent_id="support-agent") as mem:
    ctx = await mem.context_for("LangGraph issue")
    print(ctx.content)
    # → "[episodic] User is debugging a LangGraph state machine"

​
Connect to a server

To share memory across agents or enable multi-tenant isolation: 
export PLYRA_SERVER_URL=https://plyra-memory-server.politedesert-a99b9eaf.centralindia.azurecontainerapps.io
export PLYRA_API_KEY=plm_live_your_key_here
No code changes. The library detects these env vars and routes all operations to the server automatically.
Get your API key at docs.plyra.dev/memory-server/api-keys.

Memory layers

Understand working, episodic, and semantic storage.

Framework integrations

LangGraph, LangChain, AutoGen, CrewAI, and more.