@shi_weiyan: Ever watch an agent solve a problem with 100% confidence and get it completely wrong? You can fix it… if you catch it. …
Summary
New paper proposes a meta-agent that monitors AI agents and corrects their mistakes, achieving ~2x more task completion autonomously.
View Cached Full Text
Cached at: 07/03/26, 02:30 AM
Ever watch an agent solve a problem with 100% confidence and get it completely wrong?
You can fix it… if you catch it. But then you need to watch every… single… run.
New paper: We show that a meta-agent can do this for you, so your agents get ~2x more done on their own.
Meta-agents are agents that manage other agents. We built a new way to write them, in just a few lines of Python.
What does this unlock?
Agent supervisor: when your agents go off track, it rewinds their runs and gets them back on track.
CooperBench: 28.8% → 54.7%
Agent optimizer: improves your agents from their past runs.
It proposes deltas and tests them by replaying only the affected parts of the run, with 95% KV-cache reuse.
Outperforms the current best meta-optimizer (MetaHarness) by 27% on LiveCodeBench, and is 2x as fast.
Tree search for agentic RL: a meta-agent that squeezes richer signal from each training run.
It branches the run into many continuations and sees how they end. It then uses that signal to improve credit assignment in RL training.
2x GRPO’s gains on TerminalBench-2.
What makes this work? Our framework: Shepherd
It represents agents as Python functions & records runs in a Git-like history that other agents can work on.
And since meta-agents are also agents, you can stack them as high as you want!
Quickstart: http://github.com/shepherd-agents/shepherd…
Shepherd: Meta-agents can manage your agents, so you don’t have to.
Paper: http://arxiv.org/abs/2605.10913 Blog & more: http://shepherd-agents.ai
Team: @simon_ycl @dch @AnanjanN @dilarafsoylu @SunJiuding @chrmanning @shi_weiyan
@StanfordNLP @StanfordHAI × Northeastern
Many thanks to everyone whose feedback shaped this work! @lateinteraction @gneubig @MonicaSLam @ChrisGPotts @ChenSun92 @Benjamin_eecs @yi_xin_dong @PranjalAggarw16 @harshitj__ @AkshayIyer22 @ZhengxuanZenWu @ChengleiSi
And very grateful for your support @tinkerapi @modal
Similar Articles
@WenZikai: Really interesting direction for improving agent reliability. Having a meta-agent monitor and intervene when an agent g…
A new paper proposes using a meta-agent to monitor and intervene when an agent goes off track, potentially doubling autonomous task completion.
@DataScienceDojo: Most AI agents fail at the same tasks over and over. Not because the model is bad but because nobody told it how to wor…
A new paper introduces Self-Harness, a method where AI agents self-improve by analyzing their own failures, generating fixes, and testing them, leading to up to 21 percentage point improvements in pass rates.
@omarsar0: New research from Meta. (bookmark it) It's on how to fix agents that forget previously made decisions. It's well know t…
Meta researchers propose a proactive memory agent that runs alongside an unmodified action agent to surface relevant past decisions and prevent forgetting in long-horizon tasks. Their method improves pass@1 on Terminal-Bench 2.0 and tau-squared-Bench by 8.3 and 6.8 percentage points respectively.
@leanxbt: This paper completely changed how I think about how an agent fixes its own code: Generate code -> Execute it -> Explain…
This paper presents a self-debugging technique where an agent iteratively generates, executes, and explains its own code to find bugs without error messages, improving accuracy by up to 12% and matching baselines that generate 10x more candidates.
everyone's focused on whether their agent works. almost nobody asks if it's actually getting better over time
The article points out a common oversight in AI agent development: while most teams monitor task completion, few systems capture and feed failure patterns back into future runs to enable learning and improvement over time.