OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

Agentkeeper

AgentKeeper is Python-based cognitive-continuity infrastructure that stores AI agent identity, memory and priorities for reconstruction across model switches, context limits, crashes and restarts. It also supports importing ChatGPT, Claude and Gemini exports into a queryable memory view.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Agentkeeper — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use AgentKeeper when an agent must retain its stored context across model changes, context-window limits or process restarts, or when you need to inspect imported AI chat exports in one scoped memory.

Who it is for

  • Developers building long-lived AI agents that need persistent identity and memory.
  • Teams that need to checkpoint and restore agent state after a process interruption or model change.
  • People who want to bring supported ChatGPT, Claude and Gemini exports into one inspectable memory.

Documented uses

  • Preserve an agent’s identity, principles, constraints, facts and relationships across restarts.
  • Create and restore a checkpoint of cognitive state, optionally carrying an opaque execution-state payload.
  • Switch model providers while reconstructing the same stored memory in a provider-appropriate format.
  • Import and compare memory from supported AI chat exports.
Before you choose

Requirements & limits

The project is open source under the MIT license. The README documents installation from PyPI with pip and local SQLite as the default storage, with no required dependencies in the base installation.

  • The base package defaults to local SQLite and does not require external infrastructure.
  • Provider integrations are installed as optional extras; real provider use requires the corresponding credentials.
  • The README documents mock settings for a no-key, no-network demo.
  • Checkpoint restoration rebuilds stored cognitive state; the documentation does not guarantee the model’s subsequent behavior.
  • Encrypted storage, MCP access and local semantic embeddings are provided through optional installation extras.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install AgentKeeper and run its local demo

Install the package and view the documented mock-provider demonstration of identity, compression and cross-model translation.

Before you start

  • Python environment with pip
  • No API key or network connection is required for the mock demo, according to the README

Steps

  1. Install the package with the command below.

    Documented command
    pip install agentkeeper-ai

  2. Run the command below from the project repository.

    Documented command
    python examples/demo.py

Expected outcome

The README says the demo runs with mock provider and embedding settings and demonstrates identity hardening, compression, cross-model translation and async behavior.

Guide limitations

  • The mock demonstration does not require external provider credentials and is not the same as connecting a real model provider.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
Thinklanceai/agentkeeper · Official homepage title · retained in its source language
How we review sourcesSuggest a correction