It's exciting to see so much experimentation when it comes to form factors for agent orchestration!
The first question that comes to mind is: how do you think about cost control? Putting a ton in a giant context window is expensive, but unintentionally fanning out 10 agents with a slightly smaller context window is even more expensive. The answer might be "well, don't do that," and that certainly maps to the UNIX analogy, where you're given powerful and possibly destructive tools, and it's up to you to construct the workflow carefully. But I'm curious how you would approach budget when using Axe.
Great question and it's something that I've not dig into yet. But I see no problem adding a way to limit LLMs by tokens or something similar to keep the cost for the user within reason.
I've had good success with something along these lines but perhaps a bit more raw:
- claude takes a -p option
- i have a bunch of tiny scripts, each script is an agent but it only does one tiny task
- scripts can be composed in a unix pipeline
#!/usr/bin/env bash
# ai-commit-msg: stdin=git diff, stdout=conventional commit message
# Usage: git diff --staged | ai-commit-msg
set -euo pipefail
source "${AGENTS_DIR:-$HOME/.agents}/lib/agent-lib.sh"
SYSTEM=$(load_skills \
core/unix-output.md \
core/be-concise.md \
domain/git.md \
output/plain-text.md)
SYSTEM+=$'\n\nTask: Given a git diff on stdin, output a single conventional commit message. One line only.'
run_agent "$SYSTEM"
And you can see to keep the agents themselves tiny, they rely on a little lib to load the various skills and optionally apply some guard / post-exec validator. Those validators are usually simple grep or whatever to make sure there were no writes outside a given dir but sometimes they can be to enforce output correctness (always jq in my examples so far...). In theory the guard could be another claude -p call if i needed a semantic instruction.
Do you have examples of these commit messages? I have yet to see an AI write a good commit message. At least when compared to good commit messages -- if it just does better than "wip" or "fix stuff" that's not a high bar.
Context: You are working with Git repositories.
- Commit messages follow Conventional Commits: type(scope): description
- Types: feat, fix, docs, refactor, test, chore, ci, perf
- Subject line max 72 chars, imperative mood, no trailing period
- Reference issue numbers when relevant
So it produces messages like:
$ git diff HEAD~1 | bin/ai-commit-msg
fix(guards): pass input to claude and tighten verdict handling
My issue with this kind of message is that it doesn't convey intent well enough. This kind of commit message will always be like "remove check from handleUser" instead of "fix authorization in xyz case". But I assume these are different schools of commits -- I prefer commits which convey WHY, not WHAT, much like source code comments.
I'm not sure if HN is being flooded with bots or if the majority of people here nowadays lack a sense of simplicity.
Anybody looking to do interesting things should instantly ignore any project that mention "persistent memory". It speaks of scope creep or complexity obfuscation.
If a tool wants to include "persistent memory" it needs to write the 3 sentence explanation of how their scratch/notes files are piped around and what it achieves.
Not just claim "persistent memory".
I might even go so far that any project using the terminology "memory" is itself doomed to spend too much time & tokens building scaffolding for abstractions that dont work.
This is what I've been trying to get nanobot to do, so thanks for sharing this. I plan to use this for workflow definitions like filesystems.
I have a known workflow to create an RPG character with steps, lets automate some of the boilerplate by having a succession of LLMs read my preferences about each step and apply their particular pieces of data to that step of the workflow, outputting their result to successive subdirectories, so I can pub/sub the entire process and make edits to intermediate files to tweak results as I desire.
Aside but 12 MB is ... large ... for such a thing. For reference, an entire HTTP (including crypto, TLS) stack with LLM API calls in Zig would net you a binary ~400 KB on ReleaseSmall (statically linked).
You can implement an entire language, compiler, and a VM in another 500 KB (or less!)
The excessive size of Go binaries is a common complain. I last recall seeing a related discussion on Lobsters [1]. Who knows, maybe the binary could be shrunk a bit? IMHO 12mb binary size is not that big of a deal.
12 MB is not large; it's like 3 minutes of watching YouTube. Actual RAM consumption is only very weakly correlated to the binary size, and that's what matters.
It is large compared to a stripped Zig ReleaseSmall binary with no runtime. With agents, one can take this repo, and create an extremely small binary.
To your point, why even advertise the number? If that particular number is completely irrelevant in practical usage, why mention it? It seems like the point is to impress, hence my response.
This is exactly what I have wanted for a while, so thank you very much!
Disclaimer: I haven't dug into axe enough yet, just going on first impressions.
>No daemon, no GUI.
I love the world we developers live in right now. ;)
>What would you automate first?
In a sense, I have wanted to be able to just add AI to a repo, and treat it like the junior developer it is. Its okay if the junior developer will do literally any stupid thing I tell it to do, because I won't tell it to do stupid things.
So, exactly: refactor this code, implement a shim, produce docs for <blah>, construct a build harness, write unit tests, produce a build, diff these codebases, implement this API, do all this on your own branch, and build and test things so that I can review the PR over coffee.
Essentially, three word commands which will encourage the AI to produce better software. Through my repo, so I can just review through the repo.
Okay, that's how I hope things work, now off to actually dig in to axe and give it a try on a few things, thanks very much again ..
Great work! Kind of reminds me of ell (https://github.com/MadcowD/ell), which had this concept of treating prompts as small individual programs and you can pipe them together. Not sure if that particular tool is being maintained anymore, but your Axe tool caters to that audience of small short-lived composable AI agents.
> Each agent is a TOML config with a focused job. Such as code reviewer, log analyzer, commit message writer. You can run them from the CLI, pipe data in, get results out.
I'm a bit skeptical of this approach, at least for building general purpose coding agents. If the agents were humans, it would be absolutely insane to assign such fine-grained responsibilities to multiple people and ask them to collaborate.
It is easier to trust in the correctness and reliability of an LLM when you treat it as a glorified NLP function with a very narrow scope and limited responsibilities. That is to say, LLMs rarely mess up specific low level instructions, compared to open-ended, long-horizon tasks.
This is the second time I've seen somebody use the word "clankers" in the last couple days to refer to AI. Is that a thing now? Where'd that come from?
Gonna be honest, it has taken away from the message both times I've seen it. It feels a bit like you're LARPing your favorite humans vs robots tv show.
We have been rewatching Clone Wars as a family, and I, for one, find this terminology hilarious given the use of it in the series towards the separatist droids.
Really cool approach, I like the “Unix philosophy” for agents.
Curious how you handle state persistence and chaining sub-agents when agents are depth-limited.
Also, do you have any strategies for ensuring data consistency across runs, especially when multiple agents interact with the same files?
I like the idea of LLM-calling as an automation-friendly CLI tool! However, putting all my agents in ~/.config feels antithetical to this. My Bash scripts do not live there either, but rather in a separate script collection, or preferably, at their place of use (e.g. in a repo).
For example, let's say I want to add commit message generation (which I don't think is a great use of LLMs, but it is a practical example) to a repo. I would add the appropriate hook to /.git, but I would also want the agent with its instructions to live inside the repo (perhaps in an `axe` or `agents` directory).
Can Axe load agents from the current folder? Or can that be added?
A problem i have is that the agent's mental model of the system im building diverges from reality over time. After discussing that many times and asking it to remember, it becomes frustrating.
In the README you say the agents memory persists across runs, would that solve said problem?
Also, I had to do several refactorings of my agent's constructs and found out that one of them was reinventing stuff producing a plethora of function duplications: e.g. DB connection pools(i had at least four of them simultaneously).
Would AXE require shared state between chained agents? Could it do it if required?
One thing I’ve noticed when experimenting with agent pipelines is that the “single-purpose agent” model tends to make both cost control and reasoning easier. Each agent only gets the context it actually needs, which keeps prompts small and behavior easier to predict.
Where it gets interesting is when the pipeline starts producing artifacts instead of just text — reports, logs, generated files, etc. At that point the workflow starts looking less like a chat session and more like a series of composable steps producing intermediate outputs.
That’s where the Unix analogy feels particularly strong: small tools, small contexts, and explicit data flowing between steps.
Curious if you’ve experimented with workflows where agents produce artifacts (files, reports, etc.) rather than just returning text.
> Curious if you’ve experimented with workflows where agents produce artifacts (files, reports, etc.) rather than just returning text.
Yes! I run a ghost blog (a blog that does not use my name) and have axe produce artifacts. The flow is: I send the first agent a text file of my brain dump (normally spoken) which it then searched my note system for related notes, saves it to a file, then passes everything to agent 2 which make that dump a blog draft and saves it to a file, agent 3 then takes that blog draft and cleans it up to how I like it and saves it. from that point I have to take it to publish after reading and making edits myself.
That’s a really nice pipeline. The “save to file between steps” pattern seems to appear very naturally once agents start doing multi-stage work.
One thing I’ve noticed when experimenting with similar workflows is that once artifacts start accumulating (drafts, logs, intermediate reports, etc.), you start running into small infrastructure questions pretty quickly:
– where intermediate artifacts live
– how later agents reference them
– how long they should persist
– whether they’re part of the workflow state or just temporary outputs
For small pipelines the filesystem works great, but as the number of steps grows it starts to look more like a little dataflow system than just a sequence of prompts.
Do you usually just keep everything as local files, or have you experimented with something like object storage or a shared artifact layer between agents?
In my prompting framework I have a workflow that the agent would scan all the artifacts in my closed/ folder and create a yyyymmdd-archive artifact which records all artifact name and their summaries, then just delete them. Since the framework is deeply integrated with git, the artifact can be digged up from git history via the recorded names.
I have a few flows I'm using it for and have a growing list of things I want to automate. Basically, if there is a process that takes a human to do (like creating drafts or running scripts with variable data) I make axe do it.
1. I have a flow where I pass in a youtube video and the first agent calls an api to get the transcript, the second converts that transcript into a blog-like post, and the third uploads that blog-like post to instapaper.
2. Blog post drafting: I talk into my phone's notes app which gets synced via syncthing. The first agent takes that text and looks for notes in my note system for related information, than passes my raw text and notes into the next to draft a blog post, a third agent takes out all the em dashes because I'm tired of taking them out. Once that's all done then I read and edit it to be exactly what I want.
I really like seeing the movement away from MCP across the various projects. Here the composition of the new with the old (the ol' unix composability) seems to um very nicely.
OP, what have you used this on in practice, with success?
I will give it a try, I like the idea of being closer to the metal.
A Proper self-contained, self improving AI@home with the AI as the OS is my end goal, I have a nice high spec but older laptop I am currently using as a sacrificial pawn experimenting with this, but there is a big gap in my knowledge and I'm still working through GPT2 level stuff, also resources are tight when you're retired. I guess someone will get there this year the way things are going, but I'm happy to have fun until then.
This is a great concept. I fully agree with small, focused and composable design. I've been exploring a similar direction at asterai.io but focusing more on the tool layer than agent layer, with portable WASM components you write once in any language and compose together.
I currently use Claude web with an MCP component for my workflows but axe looks like it could be a nicer and quicker way to work with the tools I have.
Nice! I’ll try this soon, and I’m afraid I’ll end up using it a lot.
@jrswab, do you think it would be feasible to limit outgoing connections to a whitelist of domains, URLs, or IP addresses?
I’d like to automate some of my email, calendar, or timesheet tasks, but I’m concerned that a prompt injection could end up exfiltrating or deleting data. In fact, that’s the main reason why I’m not using Openclaw or similar projects with real data yet.
Nice approach treating LLM agents like Unix programs. The TOML config per agent is clean. I've been working on something in a similar vein for invoice processing — small focused agents that do one thing well. Curious how you handle retries when an upstream LLM provider has intermittent failures mid-pipeline?
I really like this idea. Gonna need an "Awesome Axe" page that collects agents.
One idea I'm thinking of is, after an agent has been in use for a while, and built up and understanding of the task, would be something like, "Write a Python script to replace this agent."
I could imagine this would work with agents that are processing log files or other semi-structured data for example.
Interesting approach to slimming down the framework layer. One thing I've been thinking about as agents get lighter and faster the attack surface for prompt injection and behavioral drift grows. Are you thinking about any security primitives at this layer?
This is interesting. I'd be curious to see a bunch more working examples. Personally I like the chat model because I iterate heavily on planning specs and have a lot of back and forth before implementation.
I could see using this once the plan is defined and switching back to chat while iterating on post-implementation cleanup and refactoring.
> 12MB binary, two dependencies. no framework, no Python, no Docker (unless you want it)
Does it do anything CPU-bound on its own, such that it benefits significantly from being a compiled (Go) executable? I actually like having things like this done in Python, since there's more potential to hack around with them.
If I have time I want to try this today because it matches my LLM-based work style, especially when I am using local models: I have command line tools that help me generated large one-shot prompts that I just paste into an Ollama repl - then I check back in a while.
It looks like Axe works the same way: fire off a request and later look at the results.
Nice. There's another one also written in Go (https://github.com/tbckr/sgpt), but i'll try this one too. I love that open source creates multiple solutions and you can choose the one that fits you best
Thanks! Looks like sgpt is a cool tool. Axe is oriented around automation rather than interaction like sgpt. Instead of asking something you define it once and hook it into a workflow.
This is not a replacement for either in my opinion. Apps like codex and pi are interactive but ax is non-interactive. You define an agent once and the trigger it however you please.
This looks really interesting. I'm curious to learn more about security around this project. There's a small section, but I wonder if there's more to be aware of like prompt injection
I'm happy you brought this up. I've been thinking about this and working on a plan to make it as solid as possible. For now, the best way would be to run each agent in a docker container (there is an example Dockerfile in the repo) so any destructive actions will be contained to the container.
However, this does not help if a person gives access to something like Google Calendar and a prompt tells the LLM to be destructive against that account.
looks interesting, I agree that chat is not always the right interface for agents, and a LLM boosted cli sometimes feels like the right paradigm (especially for dev related tasks).
I've not heard of that before but after looking into it I think they are solving different problems.
Dotprompt is a promt template that lives inside app code to standardize how we write prompts.
Axe is an execution runtime you run from the shell. There's no code to write (unless you want the LLM to run a script). You define the agent in TOML and run with `axe run <agent name> and pipe data into it.
Not yet but is on the short list to implement. What would you need from a session for single purpose agents? I'm seeing it more as a way to track what's been done.
Looks like an axe to me. The cutting edge of the axe is embedded into the surface. And the handle attaches near the back of the head like an axe. Most hammers I've seen the handle attaches in the middle.
Splitting mauls have a wider angle to help separate wood pieces and a beefier back to use with/as a sledgehammer or splitting wedge. What's rendered is definitely more like an axe than a splitting maul.
12MB for an "AI framework replacement"? That's either brilliant compression or someone's redefining "framework" to mean "toy model that works on my laptop." Show me the benchmarks on actual workloads, not the readme poetry.
Putting heavy AI workloads in a 12MB binary means you either make savage cuts on model support or you lock users to strange minimal formats. If you care about ops, eventually you hit edge cases where the "just works" story collapses and you end up debugging missing layers or janky hardware support. If the goal is to experiment locally or run demos, 12MB is fine but pretending it fits broader deployment is a stretch unless they're pulling some wild tricks under the hood.
It's exciting to see so much experimentation when it comes to form factors for agent orchestration!
The first question that comes to mind is: how do you think about cost control? Putting a ton in a giant context window is expensive, but unintentionally fanning out 10 agents with a slightly smaller context window is even more expensive. The answer might be "well, don't do that," and that certainly maps to the UNIX analogy, where you're given powerful and possibly destructive tools, and it's up to you to construct the workflow carefully. But I'm curious how you would approach budget when using Axe.
> how you would approach budget when using Axe
Great question and it's something that I've not dig into yet. But I see no problem adding a way to limit LLMs by tokens or something similar to keep the cost for the user within reason.
I've had good success with something along these lines but perhaps a bit more raw:
For example: Where ai-commit-msg is a tiny agent: And you can see to keep the agents themselves tiny, they rely on a little lib to load the various skills and optionally apply some guard / post-exec validator. Those validators are usually simple grep or whatever to make sure there were no writes outside a given dir but sometimes they can be to enforce output correctness (always jq in my examples so far...). In theory the guard could be another claude -p call if i needed a semantic instruction.Do you have examples of these commit messages? I have yet to see an AI write a good commit message. At least when compared to good commit messages -- if it just does better than "wip" or "fix stuff" that's not a high bar.
My skills/domain/git.md looks like this:
So it produces messages like:My issue with this kind of message is that it doesn't convey intent well enough. This kind of commit message will always be like "remove check from handleUser" instead of "fix authorization in xyz case". But I assume these are different schools of commits -- I prefer commits which convey WHY, not WHAT, much like source code comments.
I was looking at something similar. What does your agent-lib.sh look like?
Something like this: https://github.com/craigjperry2/tiny-agents/blob/main/lib/ag...
Bump.
Would love to see your tiny agents project. But understand that it might contain something sensitive and will therefore stay private.
That is probably a private repo? 404, so not something I can access
I'm not sure if HN is being flooded with bots or if the majority of people here nowadays lack a sense of simplicity.
Anybody looking to do interesting things should instantly ignore any project that mention "persistent memory". It speaks of scope creep or complexity obfuscation.
If a tool wants to include "persistent memory" it needs to write the 3 sentence explanation of how their scratch/notes files are piped around and what it achieves.
Not just claim "persistent memory".
I might even go so far that any project using the terminology "memory" is itself doomed to spend too much time & tokens building scaffolding for abstractions that dont work.
>scaffolding
The purpose of scaffolding is to create persistent memories.
>claim "persistent memory"
Just look at it as a build product.
>abstractions that don't work
Look at this as a testing problem.
This is what I've been trying to get nanobot to do, so thanks for sharing this. I plan to use this for workflow definitions like filesystems.
I have a known workflow to create an RPG character with steps, lets automate some of the boilerplate by having a succession of LLMs read my preferences about each step and apply their particular pieces of data to that step of the workflow, outputting their result to successive subdirectories, so I can pub/sub the entire process and make edits to intermediate files to tweak results as I desire.
Now that's cool!
Love to hear it! Thanks for checking it out and feel free to put up an issue on GitHub if you have any ideas for improvements.
Where is the nanobot approach not working for you?
Cool work!
Aside but 12 MB is ... large ... for such a thing. For reference, an entire HTTP (including crypto, TLS) stack with LLM API calls in Zig would net you a binary ~400 KB on ReleaseSmall (statically linked).
You can implement an entire language, compiler, and a VM in another 500 KB (or less!)
I don't think 12 MB is an impressive badge here?
it's written in golang. 12MB barely gets you "hello world" since everything is statically linked. With that in mind, the size is impressive.
golang doesn't statically link everything by default (anymore?), this is from FreeBSD:
I know off topic, but is that mostly coming from the Go runtime (how large is that about?)
The excessive size of Go binaries is a common complain. I last recall seeing a related discussion on Lobsters [1]. Who knows, maybe the binary could be shrunk a bit? IMHO 12mb binary size is not that big of a deal.
--
1: https://lobste.rs/s/tzyslr/reducing_size_go_binaries_by_up_7...
12 MB is not large; it's like 3 minutes of watching YouTube. Actual RAM consumption is only very weakly correlated to the binary size, and that's what matters.
It is large compared to a stripped Zig ReleaseSmall binary with no runtime. With agents, one can take this repo, and create an extremely small binary.
To your point, why even advertise the number? If that particular number is completely irrelevant in practical usage, why mention it? It seems like the point is to impress, hence my response.
This is exactly what I have wanted for a while, so thank you very much!
Disclaimer: I haven't dug into axe enough yet, just going on first impressions.
>No daemon, no GUI.
I love the world we developers live in right now. ;)
>What would you automate first?
In a sense, I have wanted to be able to just add AI to a repo, and treat it like the junior developer it is. Its okay if the junior developer will do literally any stupid thing I tell it to do, because I won't tell it to do stupid things.
So, exactly: refactor this code, implement a shim, produce docs for <blah>, construct a build harness, write unit tests, produce a build, diff these codebases, implement this API, do all this on your own branch, and build and test things so that I can review the PR over coffee.
Essentially, three word commands which will encourage the AI to produce better software. Through my repo, so I can just review through the repo.
Okay, that's how I hope things work, now off to actually dig in to axe and give it a try on a few things, thanks very much again ..
Reminded me of this from my bookmarks.
https://github.com/chr15m/runprompt
Great work! Kind of reminds me of ell (https://github.com/MadcowD/ell), which had this concept of treating prompts as small individual programs and you can pipe them together. Not sure if that particular tool is being maintained anymore, but your Axe tool caters to that audience of small short-lived composable AI agents.
Thanks for checking it out! And yes the tool is indeed catering to that crowed. It's a need I have and thought others could use it as well.
> Each agent is a TOML config with a focused job. Such as code reviewer, log analyzer, commit message writer. You can run them from the CLI, pipe data in, get results out.
I'm a bit skeptical of this approach, at least for building general purpose coding agents. If the agents were humans, it would be absolutely insane to assign such fine-grained responsibilities to multiple people and ask them to collaborate.
It is easier to trust in the correctness and reliability of an LLM when you treat it as a glorified NLP function with a very narrow scope and limited responsibilities. That is to say, LLMs rarely mess up specific low level instructions, compared to open-ended, long-horizon tasks.
Clankers are not humans.
This is the second time I've seen somebody use the word "clankers" in the last couple days to refer to AI. Is that a thing now? Where'd that come from?
Gonna be honest, it has taken away from the message both times I've seen it. It feels a bit like you're LARPing your favorite humans vs robots tv show.
You can find the answers to both of your questions on Wikipedia: https://en.wikipedia.org/wiki/Clanker
I've been hearing the term in IRC and discords for about a year or more already.
I get that it can seem childish but when you compare that to the indolent people who are demanding AI, it cancels out.
It mostly sounds like people who are desperate to use racist slurs and have finally found a(nother) public outlet for it.
"Clanker" is a sign that we're dealing with a Blade Runner, and better be careful
It is a thing, i've been hearing it for at least 6 months. There's a lot of people who really hate AI and want nothing to do with it.
We have been rewatching Clone Wars as a family, and I, for one, find this terminology hilarious given the use of it in the series towards the separatist droids.
I don't know exactly how these things work, but you may run into copyright/TM issues with Deque's Axe tool: https://www.deque.com/axe/devtools/
Really cool approach, I like the “Unix philosophy” for agents. Curious how you handle state persistence and chaining sub-agents when agents are depth-limited. Also, do you have any strategies for ensuring data consistency across runs, especially when multiple agents interact with the same files?
I like the idea of LLM-calling as an automation-friendly CLI tool! However, putting all my agents in ~/.config feels antithetical to this. My Bash scripts do not live there either, but rather in a separate script collection, or preferably, at their place of use (e.g. in a repo).
For example, let's say I want to add commit message generation (which I don't think is a great use of LLMs, but it is a practical example) to a repo. I would add the appropriate hook to /.git, but I would also want the agent with its instructions to live inside the repo (perhaps in an `axe` or `agents` directory).
Can Axe load agents from the current folder? Or can that be added?
Axe treats LLM agents like Unix programs—small, composable, version-controllable. Are we finally doing AI the Unix way?
That's my dream.
Dream, or _pipe_dream?
A problem i have is that the agent's mental model of the system im building diverges from reality over time. After discussing that many times and asking it to remember, it becomes frustrating. In the README you say the agents memory persists across runs, would that solve said problem?
Also, I had to do several refactorings of my agent's constructs and found out that one of them was reinventing stuff producing a plethora of function duplications: e.g. DB connection pools(i had at least four of them simultaneously).
Would AXE require shared state between chained agents? Could it do it if required?
The Unix-style framing resonates a lot.
One thing I’ve noticed when experimenting with agent pipelines is that the “single-purpose agent” model tends to make both cost control and reasoning easier. Each agent only gets the context it actually needs, which keeps prompts small and behavior easier to predict.
Where it gets interesting is when the pipeline starts producing artifacts instead of just text — reports, logs, generated files, etc. At that point the workflow starts looking less like a chat session and more like a series of composable steps producing intermediate outputs.
That’s where the Unix analogy feels particularly strong: small tools, small contexts, and explicit data flowing between steps.
Curious if you’ve experimented with workflows where agents produce artifacts (files, reports, etc.) rather than just returning text.
> Curious if you’ve experimented with workflows where agents produce artifacts (files, reports, etc.) rather than just returning text.
Yes! I run a ghost blog (a blog that does not use my name) and have axe produce artifacts. The flow is: I send the first agent a text file of my brain dump (normally spoken) which it then searched my note system for related notes, saves it to a file, then passes everything to agent 2 which make that dump a blog draft and saves it to a file, agent 3 then takes that blog draft and cleans it up to how I like it and saves it. from that point I have to take it to publish after reading and making edits myself.
That’s a really nice pipeline. The “save to file between steps” pattern seems to appear very naturally once agents start doing multi-stage work.
One thing I’ve noticed when experimenting with similar workflows is that once artifacts start accumulating (drafts, logs, intermediate reports, etc.), you start running into small infrastructure questions pretty quickly:
– where intermediate artifacts live – how later agents reference them – how long they should persist – whether they’re part of the workflow state or just temporary outputs
For small pipelines the filesystem works great, but as the number of steps grows it starts to look more like a little dataflow system than just a sequence of prompts.
Do you usually just keep everything as local files, or have you experimented with something like object storage or a shared artifact layer between agents?
In my prompting framework I have a workflow that the agent would scan all the artifacts in my closed/ folder and create a yyyymmdd-archive artifact which records all artifact name and their summaries, then just delete them. Since the framework is deeply integrated with git, the artifact can be digged up from git history via the recorded names.
What are some things you've automated using Axe?
I have a few flows I'm using it for and have a growing list of things I want to automate. Basically, if there is a process that takes a human to do (like creating drafts or running scripts with variable data) I make axe do it.
1. I have a flow where I pass in a youtube video and the first agent calls an api to get the transcript, the second converts that transcript into a blog-like post, and the third uploads that blog-like post to instapaper.
2. Blog post drafting: I talk into my phone's notes app which gets synced via syncthing. The first agent takes that text and looks for notes in my note system for related information, than passes my raw text and notes into the next to draft a blog post, a third agent takes out all the em dashes because I'm tired of taking them out. Once that's all done then I read and edit it to be exactly what I want.
great job, well done! More should be blog part, we need more seo focused.
Aren't your Hackernews answers automatised?
> - Path-sandboxed file ops. Keeps agents locked to a working directory
How is it supposed to work, if agent can simply run "cat" command instead of using skill for file read/write/etc?
chroot
you cant be serious
chroot is not a security tool and never has been
fine. cgroups, pivot_root, whatever. this is a solved problem.
I really like seeing the movement away from MCP across the various projects. Here the composition of the new with the old (the ol' unix composability) seems to um very nicely.
OP, what have you used this on in practice, with success?
I've shared a few flows I use a lot right now in some other comments.
I will give it a try, I like the idea of being closer to the metal.
A Proper self-contained, self improving AI@home with the AI as the OS is my end goal, I have a nice high spec but older laptop I am currently using as a sacrificial pawn experimenting with this, but there is a big gap in my knowledge and I'm still working through GPT2 level stuff, also resources are tight when you're retired. I guess someone will get there this year the way things are going, but I'm happy to have fun until then.
I'm excited to see how this plays out. Keep me updated on x(twitter)
This is a great concept. I fully agree with small, focused and composable design. I've been exploring a similar direction at asterai.io but focusing more on the tool layer than agent layer, with portable WASM components you write once in any language and compose together.
I currently use Claude web with an MCP component for my workflows but axe looks like it could be a nicer and quicker way to work with the tools I have.
Nice! I’ll try this soon, and I’m afraid I’ll end up using it a lot.
@jrswab, do you think it would be feasible to limit outgoing connections to a whitelist of domains, URLs, or IP addresses?
I’d like to automate some of my email, calendar, or timesheet tasks, but I’m concerned that a prompt injection could end up exfiltrating or deleting data. In fact, that’s the main reason why I’m not using Openclaw or similar projects with real data yet.
Yes, I think it will be quite trivial to make a output allow list. That's a great idea!
Nice approach treating LLM agents like Unix programs. The TOML config per agent is clean. I've been working on something in a similar vein for invoice processing — small focused agents that do one thing well. Curious how you handle retries when an upstream LLM provider has intermittent failures mid-pipeline?
I really like this idea. Gonna need an "Awesome Axe" page that collects agents.
One idea I'm thinking of is, after an agent has been in use for a while, and built up and understanding of the task, would be something like, "Write a Python script to replace this agent."
I could imagine this would work with agents that are processing log files or other semi-structured data for example.
Interesting approach to slimming down the framework layer. One thing I've been thinking about as agents get lighter and faster the attack surface for prompt injection and behavioral drift grows. Are you thinking about any security primitives at this layer?
This is interesting. I'd be curious to see a bunch more working examples. Personally I like the chat model because I iterate heavily on planning specs and have a lot of back and forth before implementation.
I could see using this once the plan is defined and switching back to chat while iterating on post-implementation cleanup and refactoring.
> 12MB binary, two dependencies. no framework, no Python, no Docker (unless you want it)
Does it do anything CPU-bound on its own, such that it benefits significantly from being a compiled (Go) executable? I actually like having things like this done in Python, since there's more potential to hack around with them.
Why not just run your typical claude/codex/pi/etc with a prompt as the command line/input?
If I have time I want to try this today because it matches my LLM-based work style, especially when I am using local models: I have command line tools that help me generated large one-shot prompts that I just paste into an Ollama repl - then I check back in a while.
It looks like Axe works the same way: fire off a request and later look at the results.
Exactly! I also made it to chain them together so each agent only gets what it needs to complete its one specific job.
I really like the project, although I would prefer a json5 config, not toml, which I find annoying to reason about.
Nice. There's another one also written in Go (https://github.com/tbckr/sgpt), but i'll try this one too. I love that open source creates multiple solutions and you can choose the one that fits you best
Thanks! Looks like sgpt is a cool tool. Axe is oriented around automation rather than interaction like sgpt. Instead of asking something you define it once and hook it into a workflow.
looks really cool, how does it differ from something like running claude headless with `claude -p`?
You don't have all the Claude Code overhead. It only gets what you give it.
what do you mean by that, not sure I understand
I’m having trouble understanding when/where I would use this? Is this a replacement for pi or codex?
This is not a replacement for either in my opinion. Apps like codex and pi are interactive but ax is non-interactive. You define an agent once and the trigger it however you please.
“ MCP support. Axe can connect any MCP server to your agents”
I just don't see this in the readme… It is not in the Features section at least.
Anyway, i have MCP server that can post inline comments into Gitlab MR. Would like to try to hook it up to the code reviewer.
Sorry, I need to update that. I just added MCP support a day or so ago.
This looks really interesting. I'm curious to learn more about security around this project. There's a small section, but I wonder if there's more to be aware of like prompt injection
I'm happy you brought this up. I've been thinking about this and working on a plan to make it as solid as possible. For now, the best way would be to run each agent in a docker container (there is an example Dockerfile in the repo) so any destructive actions will be contained to the container.
However, this does not help if a person gives access to something like Google Calendar and a prompt tells the LLM to be destructive against that account.
Ok. this is interesting. How're you handling guardrails or the agent going rouge or doing something unintended?
Now what we need is a chat interface to develop these config files.
looks interesting, I agree that chat is not always the right interface for agents, and a LLM boosted cli sometimes feels like the right paradigm (especially for dev related tasks).
how would you say this compares to similar tools like google’s dotprompt? https://google.github.io/dotprompt/getting-started/
I've not heard of that before but after looking into it I think they are solving different problems.
Dotprompt is a promt template that lives inside app code to standardize how we write prompts.
Axe is an execution runtime you run from the shell. There's no code to write (unless you want the LLM to run a script). You define the agent in TOML and run with `axe run <agent name> and pipe data into it.
There is no "session" concept?
Not yet but is on the short list to implement. What would you need from a session for single purpose agents? I'm seeing it more as a way to track what's been done.
Does it support the use of other OpenAI API compatible services like Openrouter?
Yes, I've used it with on OpenAI compatible API from an internal LLM at my job.
Thanks!
Is the axe drawing actually a hammer?
Looks like an axe to me. The cutting edge of the axe is embedded into the surface. And the handle attaches near the back of the head like an axe. Most hammers I've seen the handle attaches in the middle.
hahaha; this is what I was going for.
Just FYI, your handle is on backwards.
There are many different styles of axe and some don't flair out much.
[0]https://inchbyinch.de/wp-content/uploads/2017/08/0400-axe-ty...
[1]https://i.pinimg.com/originals/da/14/80/da148078cc1478ec6b25...
Assuming the cutting face is down, the handle is on "backwards" too (the swell at the bottom normally goes the other way).
I believe it's actually trying to render a splitting maul, which people often confuse for an axe.
Splitting mauls have a wider angle to help separate wood pieces and a beefier back to use with/as a sledgehammer or splitting wedge. What's rendered is definitely more like an axe than a splitting maul.
What you're describing is exactly what I see in the image.
Fair enough. Hard to tell one way or another with all the "action" marks.
Sure is. How weird.
Is there Gemini support?
Not yet but it will be easy to add. If you need it can you create an issue in GitHub? I should be able to get that in today.
https://github.com/jrswab/axe/blob/master/internal/refusal/r...
Lmao
Looks pretty interesting!
Tiny note: there's a typo in your repo description.
nooo! lol but thanks, I'll go hunt it down.
amazing work my friend
I LOVE IT
Why is this comment an ad?
This is the OP promoting their project — makes sense to me
How can it be an ad if it's not selling anything? Seems like a proud parent touting their child to me.
I am pretty proud of this one :)
It's a Show HN. That's the point.
Because they had an AI write it. Their other comments seem organic but the one you’re responding to does not
12MB for an "AI framework replacement"? That's either brilliant compression or someone's redefining "framework" to mean "toy model that works on my laptop." Show me the benchmarks on actual workloads, not the readme poetry.
This is not an LLM but a Binary to run LLMs as single purpose agents that can chain together.
Yeah I was disappointed by that too.
Putting heavy AI workloads in a 12MB binary means you either make savage cuts on model support or you lock users to strange minimal formats. If you care about ops, eventually you hit edge cases where the "just works" story collapses and you end up debugging missing layers or janky hardware support. If the goal is to experiment locally or run demos, 12MB is fine but pretending it fits broader deployment is a stretch unless they're pulling some wild tricks under the hood.