Until we figure out a good way to leverage humans in all of this ("Attention is all you need" applies equally to humans as it does to models) productivity gains for the system will always be limited by Amdahl's law. Gwern has an excellent post on this. [2]
why can't this be a cli tool? then you can get an agent to write a script that programmatically calls the cli tool in addition to the agent calling it directly.
It could be a cli tool, and it should be a cli tool, for exactly this reason.
Let the LLM work in code mode. Don't make it have to be the execution engine too. It can do it but it's slow and giving it tools script what it wants will go far better.
I do think there's an interesting possibility where we turn MCP into something composable. Capnproto has promise pipelining where you can issue new instructions with results you don't have yet. If MCP could copy those tricks, & express promises... and those promises worked across MCP servers ("third party handoff", https://github.com/capnproto/go-capnp/issues/597)... you'd start to have something as compellingly composable as the shell.
there is an activation energy cost to so many activities - so those things just never got done. many times it is because the cost-benefit wasn't clear at the start (unknown unknowns) so it never got done. kudos to op for experimenting and showing us one way of making something like this happen.
> Unless previously approved, we do not allow third party developers to offer Claude.ai login or rate limits for their products, including agents built on the Claude Agent SDK. Please use the API key authentication methods described in this document instead.
wow. ACP is used within zed so I guess zed is safe with ACP using claude code
I wonder if Opencode could use ACP protocol as well. ACP seems to be a good abstraction, I should probably learn more about it. Any TLDR's on how it works?
According to Opus, ACP is designed specifically for IDE clients (with coding agent “servers”), and there’s some impedance mismatch here that would need to be resolved for one agent cli to operate as a client. I havent validated this though.
—-
1. ACP Servers Expect IDE-like Clients
The ACP server interface in Claude Code is designed for:
∙ Receiving file context from an IDE
∙ Sending back edits, diagnostics, suggestions
∙ Managing a workspace-scoped session
It’s not designed for another autonomous agent to connect and say “go solve this problem for me.”
2. No Delegation/Orchestration Semantics in ACP
ACP (at least the current spec) handles:
∙ Code completions
∙ Chat interactions scoped to a workspace
∙ Tool invocations
It doesn’t have primitives for:
∙ “Here’s a task, go figure it out autonomously”
∙ Spawning sub-agents
∙ Returning when a multi-step task completes
3. Session & Context Ownership
Both tools assume they own the agentic loop. If OpenCode connects to Claude Code via ACP, who’s driving? You’d have two agents both trying to:
∙ Decide what tool to call next
∙ Maintain conversation state
∙ Handle user approval flows
Edit: I may have been rude previously but I was feeling beyond angry because I have tried to be a part of the community and I don't even know what to say
I genuinely am speechless. I think its so sad to see HN community be unable to trust and how you accuse me man which is so sad to see.
I genuinely just don't know what to say? I hope you are (jokng or something?) but what sort ofjoke is this?
Around 1980, while taking a "Saturday Morning Class" in Toronto - I discovered that there was a lab of ~24 Commodore PET 2001 (8K - blue phosphor, chiclet keyboards) at George Brown College. Spent as much time as I could there engaging with the early hacker community who all brought their shoeboxes of 5-1/4" floppies to trade programs. It was there that I had my first OMG moment when a much older kid showed me his floppy disk catalog program that could sort so much faster than mine did (he used quicksort).
What struck me when reading it was that I parachuted into that trail with zero context. So I asked an LLM to generate some surrounding context for me: this is the verbatim prompt that I just gave gpt-5.2/medium running in my own agentic harness:
> read this: https://trails.pieterma.es/trail/tempo-gradient/. then i want you to write a short essay that shows the relationship between the different quotes making sure to use each quote in the process. the narrative should help me understand the links between the different quotes in each book in context. feel free to use your own knowledge of the books and your general knowledge to make this an approachable thing. what another llm did here is cool - it found relationships between this set of quotes. now i want you to materialize them into a compelling essay that invites curiosity and teaches me something. write the essay to tempo-gradient.md and respond in chat with how you thought about the problem and whether you thought these relationships were meaningful.
This was my agent's response:
Wrote the essay to tempo-gradient.md.
How I thought about it: I treated “tempo gradient” as the shared mechanism behind all eight excerpts—mismatch in loop speed (notice →
interpret → decide → act) caused by attention, infrastructure latency, organizational rigidity, system delays, stealthy manipulation, or
runaway acceleration. Then I ordered the quotes from “inner time” (tennis/attention) → engineered milliseconds (HFT) → explicit doctrine
(OODA) → industrial schedules (mobilization) → structural delays (Beer Game/economy) → slow covert tempo (cyber) → extreme fast tempo
(AI takeoff), so each quote sets up the next.
Whether the relationships felt meaningful: yes, mostly. They’re not “the same idea” so much as the same shape appearing in different
contexts—feedback delay and decision cadence determining who gets to set the terms of reality for everyone else. The only relationship
that feels a bit more interpretive than direct is tennis ↔ geopolitics, but even there the bridge (“how many nows” as a controllable
input to decision tempo) is surprisingly sturdy.
Totally agree with this. I have seen many cases where a dumber model gets trapped in a local minima and burns a ton of tokens to escape from it (sometimes unsuccessfully). In a toy example (30 minute agentic coding session - create a markdown -> html compiler using a subset of commonmark test suite to hill climb on), dumber models would cost $18 (at retail token prices) to complete the task. Smarter models would see the trap and take only $3 to complete the task. YMMV.
Much better to look at cost per task - and good to see some benchmarks reporting this now.
For me this is sub agent usage. If I ask Claude Code to use 1-3 subagents for a task, the 5 hour limit is gone in one or two rounds. Weekly limit shortly after. They just keep producing more and more documentation about each individual intermediate step to talk to each other no matter how I edit the sub agent definitions.
Care sharing some of your sub-agent usage? I've always intended to really make use of them, but with skills, I don't know how I'd separate these in many use cases?
Had to modify them a bit, mostly taking out the parts I didn’t want them doing instead of me. Sometimes they produced good results but mostly I found that they did just as well as the main agent while being way more verbose. A task to do a big hunt or to add a backend and frontend feature using two agents at once could result in 6-8 sizable Markdown documents.
Typically I find that just adding “act as a Senior Python engineer with experience in asyncio” or some such to be nearly as good.
They're useful for context management. I use frequently for research in a codebase, looking for specific behavior, patterns, etc. That type of thing eats a lot of context because a lot of data needs to be ingested and analyzed.
If you delegate that work to a sub-agent, it does all the heavy lifting, then passes the results to the main agent. The sub-agent's context is used for all the work, not the main agent's.
Until we figure out a good way to leverage humans in all of this ("Attention is all you need" applies equally to humans as it does to models) productivity gains for the system will always be limited by Amdahl's law. Gwern has an excellent post on this. [2]
[1] https://en.wikipedia.org/wiki/Amdahl's_law [2] https://gwern.net/guardian-angel