Thinklanceai

Thinklanceai / agentkeeper

Public

Cognitive persistence layer for AI agents — cross-model memory continuity. Your agent's memory survives provider switches, crashes and restarts.

ai-agents llm-memory-python-openai-anthropic-gemini-ollama
105
12
100% credibility
Found Mar 01, 2026 at 97 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

AgentKeeper is a tool that gives AI assistants lasting memory, so they keep important facts across different AI services and restarts.

How It Works

1
🔍 Discover AgentKeeper

You learn about a helpful tool that makes AI assistants remember important details even after restarts or changes.

2
đź“Ą Set it up

You download the tool and link it to your chosen AI thinking services so everything works together.

3
🧠 Create your smart assistant

You start your personal AI helper that can hold onto memories just like a real notebook.

4
đź“ť Share key facts

You tell your assistant vital info like project budgets, deadlines, or client names, highlighting the must-know ones.

5
🔄 Ask questions anywhere

You chat with your assistant, switch thinking services anytime, and it recalls everything perfectly.

âś… Never forget again

Your AI helper stays smart and consistent forever, picking up right where it left off no matter what.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 97 to 105 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is agentkeeper?

AgentKeeper is a Python library that adds cognitive persistence to AI agents, ensuring their memory—facts marked critical or regular—survives crashes, restarts, and provider switches like OpenAI to Anthropic or Gemini. Developers store facts via simple `remember()` calls, query with `ask()`, and persist state to local SQLite, getting cross-model continuity without losing context under token limits. It tackles the stateless nature of LLMs, where agents forget everything on provider changes, delivering 95% critical fact recovery in benchmarks.

Why is it gaining traction?

Unlike workflow tools like Temporal that handle execution, AgentKeeper focuses on cognitive architecture, prioritizing facts to fight cognitive overload and github cognitive load in tight token budgets. The hook is seamless provider switching—`switch_provider("anthropic")` keeps agent's memory intact—plus stats on reconstruction efficiency, making it a lightweight cognitive tools github layer for multi-LLM setups. Early adopters value its provider-agnostic design over brittle prompt chaining.

Who should use this?

AI engineers building persistent agents for production apps, like chatbots juggling OpenAI and Ollama during outages. Teams experimenting with cognitive ai github workflows who switch models often, or indie devs prototyping cognitive science github inspired systems needing crash-proof memory. Avoid if you're locked into one provider.

Verdict

Worth a quick test for cross-model agent projects—solid API and benchmarks show real cognitive persistence meaning—but at 18 stars and 1.0% credibility score, it's early alpha with basic docs and no tests. Pair with your own validation before prod.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.