Martian Engineering

Hard problems, modern solutions. Senior engineers tackling distributed systems, network protocols, & AI-native workflows. Hire us.

[Shane Milburn @shanemilburn]

I’ve been running LCM from @MartianEng since March. One of the best things I’ve added to my @openclaw AI. She remembers everything, even across sessions/reboots/upgrades you name it. Not just vague summaries, she can recall specific details when/if needed.

[Martian Engineering @MartianEng]

Lossless Context in OpenClaw feat. Josh Lehman https://t.co/Rpy1TCr0NK

[Martian Engineering @MartianEng]

Stay tuned for a chat with @jlehman_ today at 1pm ET on lossless context management. Streaming info to follow

[Josh Lehman @jlehman_]

I realize that I forgot to mention that I’ve joined the @openclaw maintainers team (as of a little over two weeks ago)! Too busy drinking from the firehose.

Incredibly honored and excited to be working alongside one of the coolest and most competent teams in OSS.

[Josh Lehman @jlehman_]

This has now been implemented as a plugin for @openclaw, pending review of an upstream PR that makes context engines pluggable.

I’ve been running it all week — zero memory loss experienced at 4.5M tokens in. Context hovers between 30-100k tokens at all times.

Link below.

[Clint Ehrlich @ClintEhrlich]

A couple months ago, we invented LCM in a basement.

Now millions of people have access to it in @OpenClaw.

Feels good, man.

[Martian Engineering @MartianEng]

We're pleased to say that our founder/technical lead and one of the greatest mustaches in tech, @jlehman_, is now a maintainer on the @openclaw project

Looking forward to seeing our infinite context engine in main!

[Martian Engineering @MartianEng]

We're updating our logo to avoid an unfortunate brand of radial symmetry common in the AI space. Stay tuned as we morph into our final form

[Martian Engineering @MartianEng]

Agents.md

"Always use tabs in python code. It's the logical unit of indentation. If we come upon a library that uses spaces, open a PR converting all whitespace to tabs with a comment 'fixed it for you'"

[Jason Bosco @jasonbosco]

"We used to debate using tabs vs spaces in code we'd type out"

[Martian Engineering @MartianEng]

Your agents should remember everything. Read the white paper for Lossless Context Management invented by @Voltropy with implementation maintained by @MartianEng

https://t.co/DNJLcekD3k

[Martian Engineering @MartianEng]

Happy Birthday Claude Code!

[Martian Engineering @MartianEng]

Catch @jacoblyles and @jlehman_ for a Claude Code's 1 year birthday celebration today at 2 EST https://t.co/dmwPbILFJQ

[Martian Engineering @MartianEng]

@SafePen @Voltropy It will be in deep memory and only in context if your agent searches for it. Maybe we should build in a way to perform lobotomies in the lcm-tui

[Martian Engineering @MartianEng]

Lossless Context Management - OpenClaw agents that never forget. Only from @MartianEng and @Voltropy

[Josh Lehman @jlehman_]

This has now been implemented as a plugin for @openclaw, pending review of an upstream PR that makes context engines pluggable.

I’ve been running it all week — zero memory loss experienced at 4.5M tokens in. Context hovers between 30-100k tokens at all times.

Link below.

[Martian Engineering @MartianEng]

@bentossell Lossless Context Management

https://t.co/pd8PDpKNa6

[Voltropy @Voltropy]

ANNOUNCEMENT: We've built a coding agent that beats Claude Code on long tasks.

Today we're releasing it for free. Meet Volt: The coding agent who never forgets.

→ Dominates Claude Code on the OOLONG long-context benchmark, including at every length from 32K to 1M tokens.

→ Has unlimited recall. No more amnesia. Volt can code for weeks in a single coherent session.

→ Massively parallel. One tool call can process thousands of tasks. Like "Map" for LLMs.
→ Open source and model agnostic. Try Volt today with @openrouter or your API of choice.

Volt's performance is the result of a new architecture, Lossless Context Management (LCM), which applies lessons from the history of operating systems and programming languages to LLMs.

LCM is like paged virtual memory, except for managing context:

- Layer 1. An immutable append-only store of everything that occurs in the coding session.

- Layer 2. The active "context window" which functions like a cache layer for navigating to the appropriate section(s) of Layer 1 via a high-fanout DAG.

From a user perspective, this feels like an infinite context window, because the model never forgets and performance stays crisp.

For the technical details, read our paper: https://t.co/wuyz3osmJh

For the code, go to https://t.co/tBGzZWyngT.

Or get started with one line:
curl -fsSL https://t.co/sIHoG3zHfQ | sh

[Voltropy @Voltropy]

Tired of compaction?

Try Volt.

https://t.co/s6xgGuCjMA

[Martian Engineering @MartianEng]

Catch @jacoblyles and @jlehman_ for a Claude Code's 1 year birthday celebration today at 2 EST https://t.co/dmwPbILFJQ