31–40 of 140 posts
Earlier quoted context omitted.
But what is an agent without tools?
Code.
I am not sure anyone knows what a harness is at this point. I've heard 17 different definitions of it at this point. It's almost like a buzzword in search of a problem.
I don’t even know what an agent means, let alone harness.
Personally, for me it embodies a level of autonomy. I define that as, an AI model with potential to interact with something external to itself based on its output, where that includes its own future behavior.
I am not sure anyone knows what a harness is at this point. I've heard 17 different definitions of it at this point. It's almost like a buzzword in search of a problem.
AI companies would love if everything ran in their cloud, but arguably there are latency reasons or other reasons to run at least some stuff in your own computer
Personally, I find it fascinating to watch how, whenever a new technology appears, people start competing to define and own its standards. Manus rebuilt its harness five times in six months. The model stayed the same, but the architecture changed five times. LangChain re-architected Deep Research four times in one year. Anthropic also ripped out Claude Code’s agent harness whenever the model improved. Ever since Mitc…
What. The idea is as old as anyone can remember, and wrt. LLMs, it was known to be important since at least as early as ChatGPT being first released.
-What remains unsolved is what should an Agent reasonably have access to in what context and for how long (etc).
Probabilistic code that can run far faster than human driven code, we don’t have a great model yet. We all should spend our energy there…
- Separating / putting controls on the FS resource is no different than putting the agent behind a firewall / allow-deny list.
It doesn’t invalidate running a sandbox in a sandbox to have better security.
Personally, I find it fascinating to watch how, whenever a new technology appears, people start competing to define and own its standards. Manus rebuilt its harness five times in six months. The model stayed the same, but the architecture changed five times. LangChain re-architected Deep Research four times in one year. Anthropic also ripped out Claude Code’s agent harness whenever the model improved. Ever since Mitc…
> Ever since Mitchell Hashimoto mentioned the harness in February What. The idea is as old as anyone can remember, and wrt. LLMs, it was known to be important since at least as early as ChatGPT being first released.
But I think the term started being used closer to its current meaning around this point:
https://www.softwareimprovementgroup.com/blog/what-is-harnes...
In a way, the sequence was something like:
prompt engineering(23~4) -> context engineering(25) ->harness engineering(26)
At first, it was mostly understood as a correction or extension of prompt engineering. But the idea of “harness” as the layer that corrects, constrains, and operationalizes agents seems to have emerged much more clearly around 2026.
So yes, there is definitely some terminological confusion in the early phase. That is normal. New technical fields often begin with several competing names for almost the same layer, and only later does one term become stable.
I don’t get it. Calling an API requires a sandbox in most cases. The others could be abused in service of an un-sandboxed agent with API access.
If the harness is outside the sandbox then it’s just an ambiguous and confusing security model and boundary.
Personally, I find it fascinating to watch how, whenever a new technology appears, people start competing to define and own its standards. Manus rebuilt its harness five times in six months. The model stayed the same, but the architecture changed five times. LangChain re-architected Deep Research four times in one year. Anthropic also ripped out Claude Code’s agent harness whenever the model improved. Ever since Mitc…
In my opinion, the main driver here is how fast models have evolved in the past 12 months. It makes the architecture of everything around them obsolete, very fast.
We went from using models as a building block, wrapping them in heavy workflow code, to now models being smart enough to drive their own workflows and planning.
Personally, I find it fascinating to watch how, whenever a new technology appears, people start competing to define and own its standards. Manus rebuilt its harness five times in six months. The model stayed the same, but the architecture changed five times. LangChain re-architected Deep Research four times in one year. Anthropic also ripped out Claude Code’s agent harness whenever the model improved. Ever since Mitc…
Author here. In my opinion, the main driver here is how fast models have evolved in the past 12 months. It makes the architecture of everything around them obsolete, very fast. We went from using models as a building block, wrapping them in heavy workflow code, to now models being smart enough to drive their own workflows and planning.
One thing I wonder about is whether path routing alone is enough.
If `/workspace` goes to the sandbox and `/memory` or `/skills` goes to the database, the path tells you where to send the request. But it does not tell you whether this user, session, or agent is allowed to access it.
When I built something similar with an MCP filesystem, I found that I needed a scope check before actually running the operation. In my case, I was using GPT dev mode through a Cloudflare tunnel to control my local environment/model, so this kind of boundary became important.
So I like the path-routing idea, but I wonder if it eventually needs a scope or permission layer as well.
Arguably this is a feature not a bug. Conflict resolution forces the need for a process to come to agreement on a common source of truth - one of the reasons why most Git repos don’t allow users to push to main directly. Writing directly to a shared memory database seems like it would result in chaos and a host of side effects once the number of users scales.