mini-swe-agent

I think they are referring to the file https://github.com/SWE-agent/mini-swe-agent/blob/main/src/mi... 190 lines with comments included so close enough Which is the actual agent? Everything else is just glue code to connect to different LLM providers, etc

Cool study, we definitely need more principled studies on the role of harnesses. I'd also say that there aren't too many benchmarks where the more complicated harnesses consistently outperform extremely simple agents. But I'm also biased, because I wrote https://github.com/swe-agent/mini-swe-agent/ , which is probably the most minimal agent out there (it started as just 100 lines, all included), and it's used in a lot of benchmarks like DeepSWE, terminalbench, programbench (seems like it's still top of the ranking for TB3, but wasn't evaluated with the best models on TB4).

100% by better models. Since his talk models have gained more context windows (up to usable 1M), and RL (reinforcement learning) has been amazing at both picking out good traces, and taught the LLMs how to backtrack and overcome earlier wrong tokens. On top of that, RLAIF (RL with AI feedback) made earlier models better and RLVR (RL with verifiable rewards) has made them very good at both math and coding. The harnesses have helped in training the models themselves (i.e. every good trace was "baked in" the model) and have improved in enabling test time compute. But at the end of the day this is all put back into the models, and they become better. The simplest proof of this is on benchmarks like terminalbench and swe-bench with simple agents. The current top models are much better than their previous versions, when put in a loop with just

I don't think so, model improvements far outweigh any harness or tooling. Look at https://github.com/SWE-agent/mini-swe-agent for proof

We (the SWE-bench team) have a 100 line of code agent that is now pretty popular in both academic and industry labs: https://github.com/SWE-agent/mini-swe-agent I think it's a great way to dive into the agent world

> This is why we're seeing the bulk of gains from things like MCP and, now, "agents". This is objectively not true. The models have improved a ton (with data from "tools" and "agentic loops", but it's still the models that become more capable). Check out [1] a 100 LoC "LLM in a loop with just terminal access", it is now above last year's heavily harnessed SotA. > Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent! [1] - https://github.com/SWE-agent/mini-swe-agent

It's 100% the models. Terminal bench is a good indication for this. There the agents get "just a terminal tool", and yet they still can solve lots and lots of tasks. Last year you needed lots of glue, and two years ago you needed monstrosities like langchain that worked maybe once in a blue moon, if you didn't look funny at it. Check out the exercise from the swe-agent people who released a mini agent that's "terminal in a loop" and that started to get close to the engineered agents this year. https://github.com/SWE-agent/mini-swe-agent

> it shows that even if models are plateauing, The models aren't plateauing (see below). > invention of MCP was a lot more instrumental [...] than model upgrades proper Not clear. The folks at hf showed that a minimal "agentic loop" in 100 LoC [1] that gives the agent "just bash access" still got very close to SotA with all the bells and whistles (and surpassed last year models w/ handcrafted harnesses). [1] - https://github.com/SWE-agent/mini-swe-agent

Personally I found the mini SWE-agent to be a very approachable introduction to building agents: https://github.com/SWE-agent/mini-swe-agent There are a number of free and cheap LLM options to experiment with. Google offers a decent free plan for Gemini (get some extra Google accounts). Groq has a free tier including some good open weight models. There's also free endpoints on OpenRouter that are limited but might be useful for long running background agents. DeepSeek v3.2, Qwen3, Kimi K2, and GLM 4.6 are all good choices for cheap and capable models. Local models are generally not a shortcut to cheap and effective AI. It's a fun thing to explore though.

We (the Princeton SWE-bench team) have a 100 line of code agent that does pretty well, you can read the code here: https://github.com/SWE-agent/mini-swe-agent

Pretty sure you also need about 120 lines of prompting from default.yaml https://github.com/SWE-agent/mini-swe-agent/blob/7e125e5dd49...

OK that really is pretty simple, thanks for sharing. The whole thing runs on these prompts: https://github.com/SWE-agent/mini-swe-agent/blob/7e125e5dd49... Your task: {{task}}. Please reply with a single shell command in triple backticks. To finish, the first line of the output of the shell command must be 'COMPLETE_TASK_AND_SUBMIT_FINAL_OUTPUT'.

We (the Princeton SWE-bench team) built an agent in ~100 lines of code that does pretty well on SWE-bench, you might enjoy it too: https://github.com/SWE-agent/mini-swe-agent