OpenClaw Integration — Pensyve Docs

Use Pensyve as a persistent memory plugin for OpenClaw/OpenHands agents. The adapter exposes three tools — remember, recall, forget — compatible with any tool-calling agent framework.

Install

pip install pensyve

Copy the adapter into your project from the Pensyve repo:

cp integrations/openclaw/pensyve_openclaw.py your_project/

Or grab just the module:

curl -O https://raw.githubusercontent.com/major7apps/pensyve/main/integrations/openclaw/pensyve_openclaw.py

Initialize

from pensyve_openclaw import PensyvePlugin

plugin = PensyvePlugin(namespace="my-project")

Parameter	Default	Description
`namespace`	`"default"`	Isolates memories per project/agent
`path`	`~/.pensyve/`	Storage directory
`entity_name`	`"openclaw-agent"`	Name for the default agent entity

Get Tools

tools = plugin.tools  # list of 3 tool dicts

Each tool dict has name, description, parameters, and function keys — ready to pass directly to an OpenClaw/OpenHands agent or any framework that accepts tool definitions.

Tool Reference

pensyve_remember

Store a fact in persistent memory.

Parameter	Type	Required	Description
`fact`	`string`	Yes	The information to store
`confidence`	`number`	No	Confidence in [0, 1]. Default: 0.8

pensyve_recall

Search memory for relevant information. Returns memories ranked by relevance.

Parameter	Type	Required	Description
`query`	`string`	Yes	Search query
`limit`	`integer`	No	Max results. Default: 5

pensyve_forget

Clear all stored memories for the current agent entity.

Parameter	Type	Required	Description
`hard_delete`	`boolean`	No	Permanently delete instead of archiving. Default: false

Direct Usage

You can call the underlying methods directly without going through the tool interface:

from pensyve_openclaw import PensyvePlugin

plugin = PensyvePlugin(namespace="my-project")

# Store a fact
result = plugin._remember("User prefers dark mode", confidence=0.9)
print(result)  # {"status": "ok", "memory_id": "...", "content": "..."}

# Search memories
result = plugin._recall("dark mode", limit=3)
for m in result["memories"]:
    print(f"[{m['memory_type']}] {m['content']} (score={m['score']:.2f})")

# Clear memories (soft delete)
plugin._forget(hard_delete=False)

With OpenClaw Agent

Pass the tools when creating your agent:

from pensyve_openclaw import PensyvePlugin

plugin = PensyvePlugin(namespace="my-project")

# Register tools with your agent
agent = Agent(tools=plugin.tools)

The agent can now call pensyve_remember, pensyve_recall, and pensyve_forget during execution. Memories persist across sessions in the local SQLite database.

Works With Any Tool-Calling Framework

The plugin pattern is framework-agnostic. Each tool is a dict with a standard schema (name, description, parameters, function). Pensyve also ships adapters for LangChain, CrewAI, and AutoGen under integrations/.

On this page