10 months ago I would agree, but today agent means something else: a workload, and the battle is on: who will host more of those workloads and do other kind of cross sell and upsell in this environment.
"Deploy" section of that article makes the difference.
Well, I'd love to use an open-source harness! My issue is that Opus 5.5 / Fable 5.1 has been the best model for me to work with, and it's way cheaper to use it with the subscription to Claude Code Max. I won't pay more just to use an open-source harness. I'd rather just wait it out and stay on Claude Code not having to constantly jump around / switch tools as the industry matures.
You can use your claude subscription just fine in open harnesses!
I make one (https://github.com/juggler-ai/juggler) and it'll auto-detect and use your claude CLI. I use that claude plan for 90% of my work, plus a bit of GPT, I just prefer to do it inside a nicer, less locked-in environment than claude.
There's really no moat around proprietary closed harnesses, it's trivial to jump between them, but people need to be aware that it's possible!
Locked in? I routinely switch between claude and codex, same codebases, same skills, and even shared "memories" with maybe 5 minute effort to set it up. There is no lock in whatsoever.
The big hoo-har about it earlier this year was because something (opencode?) was pretending to be the claude CLI and using the API that way. Anthropic made a fuss, then GPT said "we don't mind people using their own harnesses" and everyone relaxed about it, as long as the harness goes via the claude CLI and doesn't try to cheat.
Yeah there's even an "Agent SDK" from Anthropic that manages Claude CLI sessions for you, and a separate token budget for apps that use it, on top of the subscription tokens.
even that was relaxed by Anthropic and they stopped chasing token prices for Agent SDK. I routinely use Agent SDK with my subscription and it doesn't come out of a separate budget. Their ToS vaguely says just don't use your personal subscription for commercial resale purposes.
Pi can utilize Claude or GPT subscriptions and they work wonderfully. You can have both subscriptions logged in and write your own tools to route to the model that works best in every scenario. I have access to both pro level subscriptions for work and a large credit count for api calls for Claude. I use all 3 with JEV as my router with a local model for context management and compaction. my token usage has gone down to a point where I rarely even use api calls. I would recommend oh-mi-pi for anyone who wants to get started with using an open source harness as it has very good memory and context management tools that ship with it and are extremely easy to set up.
You don't have to pay more - I wouldn't switch to an open source harness on a whim, but your harness/cheaper agents can always call Claude Code headlessly.
As context: I work as a consultant and produce code on behalf of my customers. What's most important to me is that the quality of the code is as high as it can be, without me having to rerun it with multiple different models to see which agents produces the best. If Opus 5.5/Fable 5.1 (or a different model for that sake) consistently produces the best code, I will use that for all my tasks, since that lowers the cognitive burden.
The cheaper agents never really matter, because the Max subscription costs me $100 and that is less than an hour of pay. In some cases the customer even pays for the subscription.
But I do love the idea of being able to use Pi or OpenCode. I am a heavy terminal user, and used Vim/Tmux/Lazygit-types of workflows before coding agents were in the terminal, so I do love customizability and portability. Right now there is just too much ground-shift to take part in it. Better to wait it out.
1. A lot of people already have Claude Code set up, and it's easier to keep using that than to install something new.
2. Lots of people are used to Claude Code's interface/commands/etc. and don't feel like paying the cost of learning something new. The benefits would have to outweigh those switching costs, and it's unclear that they do.
3. Claude Code already works with your Claude MAX subscription[0], which gives you an incredible amount of usage of the top-tier models for cheap, which can't be matched by any open-source options. And most people won't exhaust this usage, so they have no real need to use open-source models.
4. Although it's hard to 100% confirm this, there's reason to believe that Claude itself works better with Claude Code than with other harnesses.
5. There's no real lock-in. If tomorrow you decide you don't want to use Claude Code, you can just switch to something else ezpz. So there's no reason to worry that you're digging a hole and locking yourself in. Thus, there's no urgency to switch away.
[0] Yes, open-source harnesses can use your Claude MAX subscription, but does this "benefit" of open-source harnesses really outweigh the costs of #1, #2, and #4 above?
Yeah I thought so too, just didn't want to overstate it. So actually, a pretty big reason not to use an open-source harness if you're using a big lab's model.
I'm responding to a comment, not to article, and the comment is asking about open source harnesses, presumably in comparison to all non-open source harnesses, not just in comparison to Cursor.
A lot of the extant harnesses are built around interaction, which adds a bunch of complexity.
Eg an agent that only runs off a webhook trigger doesn’t have a use for all the UI stuff Pi does (and you may not want the local state like Pi does by default).
I’m still sort of with you, I’d generally do ACP to an open source agent. You do get much finer grained control if you use one of these agent SDKs (there’s a bajillion of these as well).
I havent switched to a 3rd party harness yet so quick question for anyone, does the claude code or codex harness not give the models a significant improvement in their ability? Or does prompting a model on openrouter vs claude code yield the same result?
There's just not really any features in an open-source harness that blow my mind enough to move off the highly-subsidized claude code subscription. I type in instructions and it spits out code. I'm increasingly of the opinion that the more complex features being floated are productivity theater.
Why would you not just build your own harness in October 2026? Gateway, memory, tool calls, LLM api - the basics are pretty simple to get started and get to a point where your harness can be self-improving to add any new features that you need.
Claude Code wont let me use their sub with one as far as I remember. So I'm stuck with their convoluted TUI. It does what I need, and when it doesn't I just build tools that Claude consumes. The new "Artifacts" feature has helped a bit though.
Anthropic employees clarified in HN comment threads that the API credits are in addition to the existing behavior and not a replacement. Other harnesses can still call the Claude Code CLI and use the subscription quota that way.
Tbh I think Thariq or one of the others passively hinted/suggested but didn't come out and make a categorical statement. I've seen anecdotal evidence of folks getting those extra data messages in various "not using the official harness" scenarios, but there's way more people like you who are not having issues.
Interesting could be a bug on my end. I'll look and see if it's a problem with how I set up my sandboxes, or something related to the precedence when both are available in the environment. With all the back and forth on their messaging at the time I took it as a cue to remove the Claude Code dependency from my workflow and start using other providers' models more.
I was both excited and disappointed to learn that Claude Code has plugins, but they're not like actual plugins to let you customize CC, which would be nice.
oh-my-pi works great with Anthropic, OpenAI and xAI subscriptions. Being able to quickly to switch between models without changing your workflow is such a joy - for me at least.
OpenCode + DeepSeek V4.1 Flash Uncensored on Venice AI is truly incredible. Granted, I have it also drive an interactive claude code session in tmux and just do everything Claude Code refuses to. But, it just gets shit done. Kind of what makes Instinct so powerful, don't tell me you won't do something, just get it done.
I wish you would explain your reasoning instead of preemptively telling people they don't know what they're talking about. Then I might actually learn something.
Aside from the opus5.5/fable + sub-lockin arguement...
Building an exceptional harness requires exceptional understanding of the technology you are harnessing.
Ant is building Claude Code in a place of frontier knowledge and advancement - everything surrounding the tool at Anthropic is something that does not exist within any "open source" organization. OC has recruited exceptional developer and new-media talent, but they have largely developed from a "catchup/copy" and designer-led process, and have consistently been behind in terms of capabilities and/or have misaligned implementation due to what I would classify as disadvantages of not understanding the underlying technology fully. I think this has led to a lot of unintended bloat - maintaining a plugin for it has been consistently very head-scratching experience.
- Pi was a refreshing welcome to the OSS arena with thoughtful design to enable simplicity and ecosystem-based development
In October 2026 - Claude Code feels further stretched ahead. The orchestration/projects model you get with sub agents either in the cloud/remote or local is the strongest capability I have experienced in awhile. Beyond the model intelligence the harness has been consistently refined and now includes things like Mods for robust customization. OC is pushing through big milestone release, not advancing on the front in terms of agentic engineering ux. Who came up with MCP, Slack/Claude Tag interface paradigm (less about slack), dynamic workflows, durable subagents, agent-to-agent communication across the harness, etc
This is my goto SDK, it comes in multiple languages too. I'm going to be wrapping the Go ADK in CUE Nov/Dec, along with Dagger, and rebuild my personal harness from scratch which I worked on last Nov/Dec.
Does anyone even use Cursor still? I loved it during the tab era but ever since they tried to catch up to agentic workflows, the product experience got worse and worse
I think their agentic cloud stuff is their best product right now. This makes sense to me. We still have subs because its nice to have something that works when Claude doesn't, it was already approved, and Composer actually is a decent value. I will sometimes use the cloud agent to put up a PR when I'm not at my computer and I remember it exists.
So a few people started using grok bot and I'll probably regret that some day, but honestly the deployment model they have here for custom agents is kind of appealing to me, much more than BigLang.
I didn't say it is THE best, just Cursor's best product - their comparative advantage. I think it makes sense for them to lean into this direction. I haven't tried Devin.
SpaceX-stans (god i can't believe i just typed that) use it, there's a whole subculture grokking (heinlein is turning in his grave) their way towards the same vibe coded nonsense as the rest of us.
I see people spending so much time creating "workflow" skills that have no benchmark whatsoever to prove they are any more effective than plain one sentence prompts.
This takes me back to the era of object oriented programming and UML diagrams. More time was spent framing the solution than implementing it. Is there maybe one more design pattern that could be applied? And when the time to implement came, all those diagrams were mostly in the way. But doing it was fun, since it allowed you to focus on the fun, big picture and skip the nitty gritty details that were actually hard to solve.
I feel like people are going down that slippery slope with harnesses and skills and what not. With the right prompt, and with the right model, and with a sprinkle of magic, the solution will emerge. Oh, it didn't? Let's see if there's a new model available, or if someone has come up with a new clever skill or something. Maybe then... Because actually solving annoying problems is not fun, even less so now that the code is a black box only the LLM can understand.
Diagrams and really, just designing the software you are trying to build is mostly about finding out painpoints before you start the programming (at least at point the time-consuming part). The diagramming allowed you to, without programming, show interactions between modules and check on a high level if you are missing edgecases that can’t be solved because of the design.
If you were just using it just to apply design patterns then you were doing it for the wrong purposes.
This is so frustrating to me. Ever heard of the planner agent reviewer agent and the tester agent framework?
I'm not sure if that's what you mean but I know many that are making their own flow like this and think that this brings in a higher level of performance..
Gotta justify that $160k/yr salary somehow I guess, and if the AI is writing all the code you used to write and your job is now to manage the AI, what better way to spend the days than making new AI tools and workflows and markdown files all day
why do we even care, this is a clear, lets move away from `people pushing software to git` to `agent is pushing software, agentic or not , to somehwere, db, git, or something else` human does involved in this unless there is a problem. JS or Python, is really not essential in this case, maybe it's WASM built from rust. it's not for human.
Cursor moving this direction, is a prep work where human no longer command that much , hence the usage will drop. e.g. their core product is vulnerable. Software factories, is the ultimate game over for our industry.
Its not bad, just most of them are created by corpos to sell slop to the masses. AI companies suck and you know it. Its not about good software, its about making as much money as possible before the hype train dies.
I have so many use cases for this. I just hope that we'll get a standard like https://www.openresponses.org/, because the idea of coupling an app to a particular vendor doesn't seem appealing at all, especially since it seems that the functionalities they offer are more or less the same. (I'm thinking about the OpenAI's Agent API and Anthropic's Agent SDK as "alternatives.)
Honest question, why shouldn't any agent SDK's API simply be a PROMPT.md with a runtime that runs the LLM with tools; why do you need excessive abstractions over that like langchain offers?
My 2c is that agents are not at the level where companies can easily just have unconstrained LLM with tools and let it be. Different parts of the workflow can have different models, different tools, different constraints, guardrails, etc.
The entrypoint still has to come from somewhere, and it's much more convenient to be able to explain the security and legal teams how data is handled, what is logged, how you can have control over a scenario where the agent does something unexpected.
Sure, for a personal project that could be enough, but we all know those companies target enterprise customers, and you would be surprised how little adoption there is when it comes to workflow automation due to the lack of controls.
This "runtime" is what you are building with LangChain.. A harness like claude code/codex etc is general purpose and its env is a shell + tools.
But with a custom runtime (basically a loop where we ask llm what to do next), we can give it specific tools which we define in code. This can help with things like authorization since, well its custom code you own.. so the runtime gets the tools scoped to the users permissions etc.
We’ve officially hit the JS framework churn phase of AI tooling.
10 months ago I would agree, but today agent means something else: a workload, and the battle is on: who will host more of those workloads and do other kind of cross sell and upsell in this environment.
"Deploy" section of that article makes the difference.
Why would you not use an open-source harness in October 2026? I am generally curious - what is the alpha?
Well, I'd love to use an open-source harness! My issue is that Opus 5.5 / Fable 5.1 has been the best model for me to work with, and it's way cheaper to use it with the subscription to Claude Code Max. I won't pay more just to use an open-source harness. I'd rather just wait it out and stay on Claude Code not having to constantly jump around / switch tools as the industry matures.
You can use your claude subscription just fine in open harnesses!
I make one (https://github.com/juggler-ai/juggler) and it'll auto-detect and use your claude CLI. I use that claude plan for 90% of my work, plus a bit of GPT, I just prefer to do it inside a nicer, less locked-in environment than claude.
There's really no moat around proprietary closed harnesses, it's trivial to jump between them, but people need to be aware that it's possible!
Locked in? I routinely switch between claude and codex, same codebases, same skills, and even shared "memories" with maybe 5 minute effort to set it up. There is no lock in whatsoever.
it goes against anthropic‘s terms of service, and people have had their Claude accounts banned for doing exactly this.
The big hoo-har about it earlier this year was because something (opencode?) was pretending to be the claude CLI and using the API that way. Anthropic made a fuss, then GPT said "we don't mind people using their own harnesses" and everyone relaxed about it, as long as the harness goes via the claude CLI and doesn't try to cheat.
Yeah there's even an "Agent SDK" from Anthropic that manages Claude CLI sessions for you, and a separate token budget for apps that use it, on top of the subscription tokens.
even that was relaxed by Anthropic and they stopped chasing token prices for Agent SDK. I routinely use Agent SDK with my subscription and it doesn't come out of a separate budget. Their ToS vaguely says just don't use your personal subscription for commercial resale purposes.
Pi can utilize Claude or GPT subscriptions and they work wonderfully. You can have both subscriptions logged in and write your own tools to route to the model that works best in every scenario. I have access to both pro level subscriptions for work and a large credit count for api calls for Claude. I use all 3 with JEV as my router with a local model for context management and compaction. my token usage has gone down to a point where I rarely even use api calls. I would recommend oh-mi-pi for anyone who wants to get started with using an open source harness as it has very good memory and context management tools that ship with it and are extremely easy to set up.
You don't have to pay more - I wouldn't switch to an open source harness on a whim, but your harness/cheaper agents can always call Claude Code headlessly.
As context: I work as a consultant and produce code on behalf of my customers. What's most important to me is that the quality of the code is as high as it can be, without me having to rerun it with multiple different models to see which agents produces the best. If Opus 5.5/Fable 5.1 (or a different model for that sake) consistently produces the best code, I will use that for all my tasks, since that lowers the cognitive burden.
The cheaper agents never really matter, because the Max subscription costs me $100 and that is less than an hour of pay. In some cases the customer even pays for the subscription.
But I do love the idea of being able to use Pi or OpenCode. I am a heavy terminal user, and used Vim/Tmux/Lazygit-types of workflows before coding agents were in the terminal, so I do love customizability and portability. Right now there is just too much ground-shift to take part in it. Better to wait it out.
But but you can get incremental benefit, possibly even higher than the switching cost!
1. A lot of people already have Claude Code set up, and it's easier to keep using that than to install something new.
2. Lots of people are used to Claude Code's interface/commands/etc. and don't feel like paying the cost of learning something new. The benefits would have to outweigh those switching costs, and it's unclear that they do.
3. Claude Code already works with your Claude MAX subscription[0], which gives you an incredible amount of usage of the top-tier models for cheap, which can't be matched by any open-source options. And most people won't exhaust this usage, so they have no real need to use open-source models.
4. Although it's hard to 100% confirm this, there's reason to believe that Claude itself works better with Claude Code than with other harnesses.
5. There's no real lock-in. If tomorrow you decide you don't want to use Claude Code, you can just switch to something else ezpz. So there's no reason to worry that you're digging a hole and locking yourself in. Thus, there's no urgency to switch away.
[0] Yes, open-source harnesses can use your Claude MAX subscription, but does this "benefit" of open-source harnesses really outweigh the costs of #1, #2, and #4 above?
>4. Although it's hard to 100% confirm this, there's reason to believe that Claude itself works better with Claude Code than with other harnesses.
I thought this was pretty much well established that all of the labs have model - harness pairings for exactly this reason
Whale Paper. https://arxiv.org/abs/2609.00196
EMNLP talk from like 2 weeks ago https://aclanthology.org/2026.findings-emnlp.2207/
Yeah I thought so too, just didn't want to overstate it. So actually, a pretty big reason not to use an open-source harness if you're using a big lab's model.
You realize this is about Cursor, not Claude Code?
I'm responding to a comment, not to article, and the comment is asking about open source harnesses, presumably in comparison to all non-open source harnesses, not just in comparison to Cursor.
A lot of the extant harnesses are built around interaction, which adds a bunch of complexity.
Eg an agent that only runs off a webhook trigger doesn’t have a use for all the UI stuff Pi does (and you may not want the local state like Pi does by default).
I’m still sort of with you, I’d generally do ACP to an open source agent. You do get much finer grained control if you use one of these agent SDKs (there’s a bajillion of these as well).
I havent switched to a 3rd party harness yet so quick question for anyone, does the claude code or codex harness not give the models a significant improvement in their ability? Or does prompting a model on openrouter vs claude code yield the same result?
There's just not really any features in an open-source harness that blow my mind enough to move off the highly-subsidized claude code subscription. I type in instructions and it spits out code. I'm increasingly of the opinion that the more complex features being floated are productivity theater.
Why would you not just build your own harness in October 2026? Gateway, memory, tool calls, LLM api - the basics are pretty simple to get started and get to a point where your harness can be self-improving to add any new features that you need.
Claude Code wont let me use their sub with one as far as I remember. So I'm stuck with their convoluted TUI. It does what I need, and when it doesn't I just build tools that Claude consumes. The new "Artifacts" feature has helped a bit though.
You can. Anthropic documents it: the Agent SDK and claude -p work signed in with your Claude plan and draw from your normal subscription limits. https://support.claude.com/en/articles/15036540-use-the-clau...
We build a desktop app on top of the SDK, so I've spent a lot of time on that page.
Not true any more, read the note at the top of that page. They give you API credits but it's limited.
Anthropic employees clarified in HN comment threads that the API credits are in addition to the existing behavior and not a replacement. Other harnesses can still call the Claude Code CLI and use the subscription quota that way.
Tbh I think Thariq or one of the others passively hinted/suggested but didn't come out and make a categorical statement. I've seen anecdotal evidence of folks getting those extra data messages in various "not using the official harness" scenarios, but there's way more people like you who are not having issues.
…have you read that note? Specifically the last half?
The wording is ambiguous but empirically when I last ran claude -p as part of a job it drew from my $200 API credit.
Can confirm it draws from my subscription
Not for me, it uses subscription quota.
Interesting could be a bug on my end. I'll look and see if it's a problem with how I set up my sandboxes, or something related to the precedence when both are available in the environment. With all the back and forth on their messaging at the time I took it as a cue to remove the Claude Code dependency from my workflow and start using other providers' models more.
I can confirm that the priority is alwys given to credentials, as they themselves state in the documentation they updated yesterday: https://support.claude.com/en/articles/15036540-use-the-clau... Essentially, in the -p option, if an ANTHROPIC_API_KEY is present, it is always used instead of the login; see here https://code.claude.com/docs/en/authentication#authenticatio..., so I would suggest you check your environment variables
It's the one features that keeps me using claude, it's simply convenient.
I was both excited and disappointed to learn that Claude Code has plugins, but they're not like actual plugins to let you customize CC, which would be nice.
oh-my-pi works great with Anthropic, OpenAI and xAI subscriptions. Being able to quickly to switch between models without changing your workflow is such a joy - for me at least.
Cheaper tokens in the $200 subscription.
Correct me if I'm wrong, but using Pi etc you're paying the full API price.
Only if you use Claude via /login in pi... in which case
;)You're wrong.
better discounts when compare with API cost, and frontier lab harnesses and models are tuned to each other
A main criteria in 2026 should be whether the harness gives you access to all model families for just the harness author's models.
More and more harnesses are headed this way, there are two notable holdouts.
If you tell me I should OpenCode over Claude Code in October 2026 - then you have no idea what you are talking about.
OpenCode + DeepSeek V4.1 Flash Uncensored on Venice AI is truly incredible. Granted, I have it also drive an interactive claude code session in tmux and just do everything Claude Code refuses to. But, it just gets shit done. Kind of what makes Instinct so powerful, don't tell me you won't do something, just get it done.
Interesting on Instinct, I have to often tell it "just do it"
I wish you would explain your reasoning instead of preemptively telling people they don't know what they're talking about. Then I might actually learn something.
Aside from the opus5.5/fable + sub-lockin arguement...
Building an exceptional harness requires exceptional understanding of the technology you are harnessing.
Ant is building Claude Code in a place of frontier knowledge and advancement - everything surrounding the tool at Anthropic is something that does not exist within any "open source" organization. OC has recruited exceptional developer and new-media talent, but they have largely developed from a "catchup/copy" and designer-led process, and have consistently been behind in terms of capabilities and/or have misaligned implementation due to what I would classify as disadvantages of not understanding the underlying technology fully. I think this has led to a lot of unintended bloat - maintaining a plugin for it has been consistently very head-scratching experience.
- Pi was a refreshing welcome to the OSS arena with thoughtful design to enable simplicity and ecosystem-based development
In October 2026 - Claude Code feels further stretched ahead. The orchestration/projects model you get with sub agents either in the cloud/remote or local is the strongest capability I have experienced in awhile. Beyond the model intelligence the harness has been consistently refined and now includes things like Mods for robust customization. OC is pushing through big milestone release, not advancing on the front in terms of agentic engineering ux. Who came up with MCP, Slack/Claude Tag interface paradigm (less about slack), dynamic workflows, durable subagents, agent-to-agent communication across the harness, etc
What if I'm trying to bypass the security monitor built-into CC?
What's your reasoning? What makes Claude Code better? Genuinely curious to learn.
say more, what do you think CC does better?
Im not sure how they would know, they're clearly using a singular harness.
I actively use all harnesses and maintain popular plugins across them
> I actively use all harnesses
Several perhaps, but you definitely aren't using all coding harnesses actively.
What about an AgentFactoryFactory? That would increase Cursor's valuation from $60 to $120 billion.
An IQ too large to measure
Open AI and Anthropic already have "Agent SDK" brands. Vercel is "AI SDK". I wish folks would come up with some good names for new projects.
And Google has ADK (Agent Development Kit)
Temporal workflows are my main building block for supporting agents working in multiple steps.
This is my goto SDK, it comes in multiple languages too. I'm going to be wrapping the Go ADK in CUE Nov/Dec, along with Dagger, and rebuild my personal harness from scratch which I worked on last Nov/Dec.
ADK has a graph/workflow concept too: https://adk.dev/graphs/
Are you this guy? https://github.com/hofstadter-io/hof/tree/_next Third time I've seen CUE mentioned in this context in as many days.
Not to be confused with ACP (Agent Client Protocol).
Not to be confused with ACP (Agent Communication Protocol).
Does anyone even use Cursor still? I loved it during the tab era but ever since they tried to catch up to agentic workflows, the product experience got worse and worse
I think their agentic cloud stuff is their best product right now. This makes sense to me. We still have subs because its nice to have something that works when Claude doesn't, it was already approved, and Composer actually is a decent value. I will sometimes use the cloud agent to put up a PR when I'm not at my computer and I remember it exists.
So a few people started using grok bot and I'll probably regret that some day, but honestly the deployment model they have here for custom agents is kind of appealing to me, much more than BigLang.
why is it the best? what's the differentiation to Devin?
I didn't say it is THE best, just Cursor's best product - their comparative advantage. I think it makes sense for them to lean into this direction. I haven't tried Devin.
SpaceX-stans (god i can't believe i just typed that) use it, there's a whole subculture grokking (heinlein is turning in his grave) their way towards the same vibe coded nonsense as the rest of us.
I’ve been using Vercel’s Eve it’s really lovely for an agent that lives on a server somewhere (no affiliation)
In 2026, the wheel that keeps getting reinvented is the agent harness.
If you are on onstagram 99% of posts are people shilling skills where you need to subscribe to some service. To which they are an affiliate.
And a lot of the apps are the same thing.
It's going to be a race to the bottom.
I'm starting to think that the LLMs ecosystem is going in the JS ecosystem direction, endless technologies with little to no adaption.
Literally everyday there is a new tool and the tools released last week are deprecated.
I see people spending so much time creating "workflow" skills that have no benchmark whatsoever to prove they are any more effective than plain one sentence prompts.
This takes me back to the era of object oriented programming and UML diagrams. More time was spent framing the solution than implementing it. Is there maybe one more design pattern that could be applied? And when the time to implement came, all those diagrams were mostly in the way. But doing it was fun, since it allowed you to focus on the fun, big picture and skip the nitty gritty details that were actually hard to solve.
I feel like people are going down that slippery slope with harnesses and skills and what not. With the right prompt, and with the right model, and with a sprinkle of magic, the solution will emerge. Oh, it didn't? Let's see if there's a new model available, or if someone has come up with a new clever skill or something. Maybe then... Because actually solving annoying problems is not fun, even less so now that the code is a black box only the LLM can understand.
Diagrams and really, just designing the software you are trying to build is mostly about finding out painpoints before you start the programming (at least at point the time-consuming part). The diagramming allowed you to, without programming, show interactions between modules and check on a high level if you are missing edgecases that can’t be solved because of the design. If you were just using it just to apply design patterns then you were doing it for the wrong purposes.
I still do diagrams. Not UML specifically, but using mermaid.js. It is a great way to think about the requirements before coding
This 1000x.
At the same time, the lack of this makes it then very easy for me to tune out harness noise.
It reminds me of people customizing their nvim configs all day, thinking they're being productive but not getting all that much done.
I spent a solid week on mine and haven’t had to tweak it in three years. We’re not all inefficient ;)
Hey I never deluded myself into thinking I was being productive :)
This is so frustrating to me. Ever heard of the planner agent reviewer agent and the tester agent framework?
I'm not sure if that's what you mean but I know many that are making their own flow like this and think that this brings in a higher level of performance..
> ... think that this brings in a higher level of performance.
Sure, but how are you measuring this?
I would imagine that performance is coming from essentially dedicating 3x context windows to one task, rather than any particular inter-agent magic.
I mean... isn't that the point of having multiple agents? You can more easily manipulate context to get more consistent results.
Gotta justify that $160k/yr salary somehow I guess, and if the AI is writing all the code you used to write and your job is now to manage the AI, what better way to spend the days than making new AI tools and workflows and markdown files all day
That's the whole point with AI. It's consumerism applied to software... And it sucks
why do we even care, this is a clear, lets move away from `people pushing software to git` to `agent is pushing software, agentic or not , to somehwere, db, git, or something else` human does involved in this unless there is a problem. JS or Python, is really not essential in this case, maybe it's WASM built from rust. it's not for human.
Cursor moving this direction, is a prep work where human no longer command that much , hence the usage will drop. e.g. their core product is vulnerable. Software factories, is the ultimate game over for our industry.
I attribute this to "it's so new and we don't know what the right answer is yet", but you may very well be right
> endless technologies with little to no adaption.
I think the difference is that those endless technologies will probably be used by other agents without people's explicit knowledge.
I don't think I'll ever understand this argument. Especially from purported "hackers."
Why is a large number of ideas, technologies, frameworks, etc. bad?
Hackers hate frameworks because they hide the underlying truth of a system.
'coz it's pointless if you can get the same results without them.
Its not bad, just most of them are created by corpos to sell slop to the masses. AI companies suck and you know it. Its not about good software, its about making as much money as possible before the hype train dies.
I have so many use cases for this. I just hope that we'll get a standard like https://www.openresponses.org/, because the idea of coupling an app to a particular vendor doesn't seem appealing at all, especially since it seems that the functionalities they offer are more or less the same. (I'm thinking about the OpenAI's Agent API and Anthropic's Agent SDK as "alternatives.)
Pi durable for me
Also see the existing Cursor Typescript SDK.
https://cursor.com/blog/typescript-sdk
The video is still showing @grok/agent-sdk.
Meanwhile, we know Grok Bot was really Cursor beforehand... it must be wild in there.
...why? params: [ { id: "effort", value: "high" }, { id: "fast", value: "true" }, ],
Why would anyone use this over the Anthropic sdk
You mean except for all the obvious reasons?
Honest question, why shouldn't any agent SDK's API simply be a PROMPT.md with a runtime that runs the LLM with tools; why do you need excessive abstractions over that like langchain offers?
Is it to have boundaries of some determinism?
My 2c is that agents are not at the level where companies can easily just have unconstrained LLM with tools and let it be. Different parts of the workflow can have different models, different tools, different constraints, guardrails, etc.
The entrypoint still has to come from somewhere, and it's much more convenient to be able to explain the security and legal teams how data is handled, what is logged, how you can have control over a scenario where the agent does something unexpected.
Sure, for a personal project that could be enough, but we all know those companies target enterprise customers, and you would be surprised how little adoption there is when it comes to workflow automation due to the lack of controls.
What do you use for that?
>runtime that runs the LLM with tools
This "runtime" is what you are building with LangChain.. A harness like claude code/codex etc is general purpose and its env is a shell + tools. But with a custom runtime (basically a loop where we ask llm what to do next), we can give it specific tools which we define in code. This can help with things like authorization since, well its custom code you own.. so the runtime gets the tools scoped to the users permissions etc.
What's the point in putting the vibe back in if you're implementing a specific software development process to avoid the vibe?