I have been interested in long-horizon coding tasks for a while, especially with benchmarks like FrontierSWE, where even the best coding agents like Codex and Claude Code struggle to complete tasks.

These agents come with a collection of tools like bash, file edits, grep, glob, etc.

Lazarus takes a different approach. The idea is to give the model exactly one tool: a persistent Python runtime.

Model writes Python code, executes it, and receives stdout/stderr. Through Python it inspects repos, reads and edits files, runs builds, executes tests, invokes linters, even build custom harnesses and automate whatever workflows it needs.

The motivation for this was: - Tool selection itself is a planning problem.

- Specialized tools are often difficult to compose together efficiently.

- Long-horizon tasks frequently require custom workflows that predefined tools don't provide.

- Python is expressive enough for the model to build those workflows itself.

Another decision is avoid agent hierarchies. Lazarus runs a single tool-calling loop rather than managers, planners, and worker agents.

The intuition being current models are much better at writing code than coordinating fleets of agents. Agent orchestration consumes context, introduces extra modes of failure, and adds complexity.

How does Lazarus manage context? When the "usable" context window of a model is nearly exhausted, the model gets one final opportunity to execute a Python tool call, containing anything it wants to preserve: notes, plans, functions, summaries, partial results, etc.

The loop is then restarted with only:

- The original user task

- The carryover cell

- The carryover cell's output

This allows the agent to periodically compress its own state and continue working without requiring an ever-growing context window.

I evaluated Lazarus on two FrontierSWE tasks: - git-to-zig (rewriting git in zig) - dart-style-haskell (rewriting dart-style formatter in haskell)

The runs with scores are available here: https://github.com/ExpressGradient/frontier-swe-lazarus-runs

Using GPT-5.5 at medium reasoning effort, Lazarus achieved scores comparable to reported GPT-5.5 in Codex with xhigh reasoning.

The runs were not completed to exhaustion, I stopped them because I ran out of OpenAI credits. So I suspect there is still room for improvement from longer runtimes and higher reasoning.

The project is still early, but the results made me wonder whether coding agents have become over-specialized around tool collections and orchestration systems, while under-investing in giving models a programmable environment they can shape themselves.

Lazarus: https://github.com/ExpressGradient/lazarus

What people don't get about safety at Anthropic

How Elon Musk Killed Hundreds of Thousands of People

S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic

How to Stop Shipping Low-Quality RL Environments (With Examples)

Show HN: Busbar – every LLM behind one URL, in a single Rust binary

UK orders Google to allow publishers to opt out of AI scraping

The effects of foods on LDL cholesterol levels

I Put ChatGPT Browser Inside My Terminal [video]

The Wrath of the Killdozer (2009)

Data Centers Have a New Adversary: Tigers and Leopards at a Zoo

Amazon Employees Show Up to City Council Meetings, Demand Limits on Data Centers

We Built Plainform and What It Means for Your Next Project

Transformers Are Inherently Succinct

Jax Back Ends and Devices

Tech sovereignty package to strengthen Europe's digital autonomy and resilience

Show HN: SupXML, modern memory-safe XML parser replacement for libxml2

Against an Increasingly User-Hostile Web (2017)

Pasteur, a zero-knowledge pastebin as an unikernel in OCaml

Employees aren't resisting AI – they're resisting fear

OpenClaw Got Safer in Public

Digital Dead Man's Switch for Your Files

What is my IP address?

Show HN: Lazarus, a coding agent for long-horizon tasks

Are Memories Transferable – Or Edible?

AI enthusiasts race against time, AI skeptics race against entropy

Why Can't California Count?

Neocities domain suspended by Namecheap for unrelated court case

The Fitbit Air is a good wearable weighed down by a chatty AI "coach"

Assessing the Effect of a Deep-Rooted Grass on Belowground Carbon Storage

How Not to Die (2007)

Show HN: Interact with your .eml files using MCP tools