Skillget

lerim-cli

agent
Security Audit
Pass
Health Pass
  • License — License: NOASSERTION
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 66 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as a background memory agent for coding workflows. It watches your coding-agent sessions (like Claude Code or Cursor) to automatically extract, consolidate, and store reusable project context and decisions locally.

Security Assessment
Overall Risk: Low. The application operates entirely locally and does not request dangerous permissions. It naturally accesses sensitive data—such as your source code and coding session histories—to perform its job. However, a light code scan of 12 files found no dangerous patterns, hardcoded secrets, or suspicious network requests. The extracted memories are safely stored as plain markdown files within your local project directory.

Quality Assessment
The project is actively maintained and recently updated. It has garnered 66 GitHub stars, indicating a fair level of community trust and early adoption. While the automated audit check notes the license as "NOASSERTION," the repository's documentation clearly states it uses the BSL-1.1 (Business Source License). Developers should be aware that BSL-1.1 is a source-available license that applies specific usage restrictions, meaning it is not a traditional open-source license like MIT.

Verdict
Safe to use, but review the BSL-1.1 license terms to ensure compliance with your project's needs.
SUMMARY

Background memory agent for coding workflows. Automatically extracts reusable project memory from coding-agent sessions.

README.md

Lerim Logo

Background memory agent for coding workflows.

PyPI version Python versions License Tests GitHub stars

lerim.dev · docs · pypi

Lerim

Lerim watches coding-agent sessions and builds reusable project memory automatically.

It helps your coding workflow keep memory across sessions and across tools, without vendor lock-in. Instead of losing decisions, reasoning, and project context every time a session ends, Lerim extracts and consolidates that memory in the background and stores it locally as plain markdown.

Supported session adapters today: Claude Code, Codex CLI, Cursor, and OpenCode.

Why Lerim

Coding agents are useful, but they forget too much.

A typical workflow looks like this:

  • you work with an agent
  • important decisions get made
  • the session ends
  • the next session starts with less context
  • the same reasoning gets repeated again

Lerim fixes that.

It runs as a background memory agent for coding workflows. It watches sessions, extracts durable project memory, consolidates it over time, and lets you inspect or query what the workflow has learned.

What makes Lerim different

Many tools give you memory infrastructure.

Lerim is different because it is workflow-native.

It does not only store memory.
It actively works on your coding workflow.

Lerim is built around three jobs:

  1. Extract memory from coding-agent sessions
  2. Consolidate memory over time
  3. Track project stream status as work evolves

That means Lerim is not just a database, vector store, or memory SDK.

It is a background memory agent.

What you get

With Lerim, you can:

  • keep project decisions across sessions
  • preserve reasoning and implementation context
  • share memory across different coding agents
  • ask questions against past work
  • keep memory local and file-based

Memories are stored as plain markdown in:

<repo>/.lerim/memory/

with fallback storage under:

~/.lerim/memory/

Quick start

Prerequisites:

  • Python 3.10+
  • Docker recommended

Install Lerim:

pip install lerim

Start the service:

lerim up

Check that it is running:

lerim status

Or watch live activity:

lerim status --live

What the commands do

lerim up

Starts Lerim in the background.

This is the command you run when you want Lerim to begin watching your workflow and processing memory tasks.

lerim status

Shows service health and current status.

Useful for checking whether Lerim is up and connected.

lerim status --live

Shows live status updates.

This is the best command for demos because it makes background extraction visible.

lerim sync

Indexes sessions and extracts candidate memories from recent work. This is done automatically by the docker container and based on the intervals you set in the config file under ~/.lerim/config.toml file.

lerim maintain

Improves memory quality over time by merging duplicates, archiving weak items, and refreshing useful memories. This will also runs based on the interval you set in the ~/.lerim/config.toml file.

lerim ask

Lets you ask questions against accumulated project memory.

Example:

lerim ask "Why did we choose SQLite for local metadata?"

Configuration

lerim init can help with setup. Then you can override the configs in the ~/.lerim/config.toml file.

API keys are read from environment variables, stored by default in:

~/.lerim/.env

Example:

MINIMAX_API_KEY=your-key
OPENROUTER_API_KEY=your-key
OPENAI_API_KEY=your-key
ZAI_API_KEY=your-key

Example provider config:

[roles.agent]
provider = "minimax"
model = "MiniMax-M2.7"
fallback_models = ["zai:glm-4.7"]

Most-used commands

lerim status
lerim status --live
lerim logs --follow
lerim queue
lerim queue --failed
lerim unscoped --limit 20
lerim memory list --limit 20

Setup and management:

lerim connect auto
lerim project list
lerim project remove <name>
lerim skill install

Alternative to Docker:

lerim serve

How Lerim works

Lerim runs three agent flows:

  • sync for indexing sessions and extracting memories
  • maintain for improving memory quality over time
  • ask for answering questions with memory context and citations

This makes Lerim useful not only as storage, but as an ongoing background process for project memory.

Who Lerim is for

Lerim is for developers who:

  • use coding agents regularly
  • work across multiple sessions
  • switch between different coding tools
  • want local, reusable, project-level memory
  • want memory continuity without vendor lock-in

What Lerim is not

Lerim is not just a vector store.

Lerim is not only a memory SDK.

Lerim is not tied to one coding assistant.

It is a background memory agent for coding workflows.

Docs

Development

uv venv && source .venv/bin/activate
uv pip install -e '.[test]'
tests/run_tests.sh unit
tests/run_tests.sh quality

Contributing

Contributions are welcome.

If you want to help, good starting points are:

  • session adapters and adding more agents

  • extraction quality

  • memory consolidation quality

  • docs and demo examples

  • Read the Contributing Guide

  • Browse open issues

  • Agent adapter PRs are especially appreciated -- see src/lerim/adapters/ for examples

Reviews (0)

No results found