lerim-dev

lerim-dev / lerim-cli

Public

Continual learning layer for coding agents

41
4
100% credibility
Found Mar 04, 2026 at 41 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Lerim captures and shares knowledge from AI coding agent sessions across tools like Claude Code and Cursor by storing decisions and learnings as human-readable markdown files in your project.

How It Works

1
🔍 Discover the forgetting problem

You notice your AI coding helpers like Claude or Cursor forget everything between chats, wasting time re-explaining.

2
📦 Get Lerim with one command

Install it simply so it starts watching your helpers' chats automatically.

3
🧙 Run the friendly setup

Follow the quick guide to connect your projects and helpers.

4
🚀 Turn it on

Click to launch and see your first memories appear as you code.

5
📚 Share the skill with helpers

Copy a note so your AI friends know to check past wisdom.

💡 Never explain twice

Ask questions about past choices or let helpers recall instantly, speeding up your work forever.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 41 to 41 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 lerim-cli?

Lerim-cli is a Python-based continual learning layer that adds persistent, shared memory to coding agents like Claude Code, Cursor, Codex CLI, and OpenCode. It watches agent sessions across platforms, automatically extracts decisions and learnings using LLM pipelines, and stores them as plain markdown files in your repo's `.lerim/` folder—tackling agent context amnesia and catastrophic forgetting head-on. Run `lerim up` for a Docker service with sync, maintain, and a local dashboard at localhost:8765; query memories via `lerim ask "why Postgres?"`.

Why is it gaining traction?

In a world of siloed AI agents, lerim unifies knowledge across sessions and tools—no more re-explaining patterns or failures. Its file-first approach keeps memories human-readable without databases, while continual refinement merges duplicates and applies decay for relevance. Developers dig the CLI skills for agents, graph explorer dashboard, and evals framework benchmarking extraction quality against coding agent judges.

Who should use this?

Backend devs juggling Claude and Cursor on sprawling repos, where repeating "we ditched Redis for X" kills flow. AI agent power users building continual learning llm github workflows, fighting neural network continual learning pitfalls like forgetting. Teams evaluating github continual learning tools before scaling to continual pretraining github or continual reinforcement learning github setups.

Verdict

Try it if you're deep in multi-agent coding—solid docs, dashboard, and evals make it production-ready despite 41 stars and 1.0% credibility score signaling early maturity. PRs for new adapters could accelerate adoption.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.