Blindsight (and the excellent sequel, Echopraxia) is indeed great.
Solaris by Lem is perhaps the one above all. Lem wrote several of these "inscrutable alien first contact" novels: His Master's Voice, The Invincible, Fiasco, and Eden are basically all variations on this theme, each one unique and highlighting a different aspect of humans' inability to understand the universe. The last three are a little dated now, but still enjoyable to read. HMV is rather dry, a Borgesian essay on an investigation into an alien signal, with lots of references to fictional scientific papers. (Len also wrote two collections of very Borgesian essays that are basically reviews of fictional books: A Perfect Vacuum and Imaginary Magnitude. They're interesting and funny, but I wouldn't put them among his most entertaining work.)
Roadside Picnic by Arkady and Boris Strugatsky is also a masterpiece. They also have a few stories about unseen aliens manipulating the history of humanity by placing traps or transforming humans into infiltrators. The Max Kammerer books (e.g. Beetle in the Anthill) involve this storyline and are very good, probably not well known today.
I tried Tchaikovsky (both Children of Time and Shroud) and found him to be completely unengaging as a writer. Just really dull writing and flat characters. Watts and Reynolds are much better writers. Watts in particular can really pack a punch.
OP here. Sorry if this was premature. I came across it through your earlier comment on HN, started using it (as did a colleague), and we've been impressed enough with how efficient it is that I decided it deserved a post!
I've seen sandbox policy documents for agents before, but this is the first ready-to-use app I've come across.
I've only had a couple of points of friction so far:
- Files like .gitconfig and .gitignore in the home folder aren't accessible, and can't be made accessible without granting read only access to the home folder, I think?
- Process access is limited, so I can't ask Claude to run lldb or pkill or other commands that can help me debug local processes.
For handling global rules (like ~/.gitconfig and ~/.gitignore), I keep a local policy file that whitelists my "shared globals" paths, and I tell Safehouse to include that policy by default. I just updated the README with an example that might be useful[1]. I also enabled access to ~/.gitignore by default as it's a common enough default.
For process management, there is a blurry line about how much to allow without undermining the sandboxing concept. I just added new integrations[2] to allow more process control and lldb, but I don't know this area well. You can try cloning the repo, asking your agents to tweak the rules in the repo until your use-case works, and send a PR - I'll merge it!
Alternatively, using the "custom policy" feature above, you can selectively grant broad access to your tools (you can use log monitoring to see rejections, and then add more permisions into the policy file)
That is very useful. I wasn't sure if I could supply my own override list or how I would even format one, but this solves that problem!
The process control policy, that's kind of niche and should definitely not be something agents are always allowed to do, so having a shorthand flag like you added in that pull request is the right choice.
I'm sure Anthropic and the other major players will catch up and add better sandboxing eventually, but for now, this tool has been exactly what I needed — many thanks!
I also wonder if this could have be a plugin or MCP server? I was using this plugin [1] for a bit, and it appears to use a "PreToolUse" that modifies every tool invocation. The benefit here would be that you could even change the Safehouse settings inside a session, e.g. turn process control on or off.
This would be slash commands that the agent itself wouldn't be able to do, and which would communicate with the plugin via a side channel the agent wouldn't know about. Admittedly I don't know much about the plugin interface in Claude Code, though.
When you say clench, do you mean clench the muscles (i.e. as if biting down), or do you mean jutting your jaw forward?
There is a well known phenomenon among people with (at least some types of) tinnitus that moving the jaw forward increases the sound, but that this also makes the tinnitus go away for a bit. The way my ENT explained it, it has to do with how your brain calibrates sound. Pushing the jaw forward makes the sound louder, which also causes your brain to adjust your hearing to be less sensitive. Or something like this.
With some types of tinnitus, there is a specific connection to the temporomandibular joint. My understanding is that the causes tinnitus are poorly understood, however. There are many hypotheses, but little solid evidence.
I really wish there was a seamless system for this. Once you try to do this kind of thing, you run into all sorts of rabbit holes and cans of worms.
For example, coalescing blobs into "superblobs" to avoid a proliferation of small objects means you invent a whole system for tracking "subfiles" within a bigger file.
And you'll need a compacting job to ensure old, deleted data is expunged, which may be more important than you think if the data has to be erased for privacy or legal reasons.
Object storage has no in-place mutation, so this compaction has to be transactionally safe and must be careful not to leave behind cruft on failure, and so on.
Furthermore, storing blobs in object storage without keeping a local inventory of them is, in my experience, a disaster. For example, if your database has tenants or some other structural grouping, something simple like finding out how much blob storage a specific tenant has is a very time-consuming operation on S3/GCS/etc. because you need to filter the whole bucket by prefix. So for every blob you store, you want to have a database table of what they are so that the only object operations you do are reads and writes, not metadata operations.
Sure, you have things like inventory reports on GCS that can help, but I would still say that you need to track this stuff transactionally. The database must be the source of truth, and the object storage must never be used as a database.
And so on.
This need to be able to store many small objects in object storage is coming up more and more for me, as is the desire to mutate them in-place or at least append. For example, imagine you want to build a kind of database which stores a replicated copy of itself in the cloud. There is no way to do this in S3-like object storage without representing this as a series of immutable "snapshots" and "deltas". It's fast to append this way, but you run into the problem of eventually needing to compact, and you absolutely have to batch up the uploads in order to avoid writing too many small objects.
So lately I've pondered using something else for this type of work, like a key/value database, like FoundationDB or TiKV, or even something like Ceph. I wonder if anyone else has tried that?
Well, I think this is what our company, Archil, is working on. We basically built an SSD clustering layer that proxies/caches/and assembles requests into object storage so that you can run a POSIX file system directly on top.
There's also some really great projects like SlateDB in this space, which could be more like what you're looking for (~RocksDB like API that runs on S3).
Well we have made small objects work well on Tigris (https://www.tigrisdata.com/). And we have several use cases of folks using it as KV store. Funny that you mention FoundationDB, we use that for our metadata storage.
I've heard good things about Tigris. If that means I can store billions of objects without being bankrupted by request cost, and it has fast read access (GCS is quite poor here), then that helps a lot! I'm looking right now for a system that lets me store lots of very small blobs around 4KB each.
Lots of engineers write, too. My favorite example of "very different day job" is Gene Wolfe, who worked as an industrial engineer (he famously worked on the original machine that produced Pringles potato chips), and spent most of his working life as an editor of the magazine Plant Engineering.
R. A. Lafferty worked as a full-time electrical engineer for Clark Electrical Supply Company all his life, though he eventually moved to a salesperson position.
You're the second commentor I've read mentioning Gene Wolfe, in the past two days [0]. After I finish my Cormick McCarthy stint, I'll start his The Book of the New Sun this spring.
Vonnegut is among my favorite authors alive during my lifetime — he was a POW during the bombing of Dresden (WWII) — a great drafted veteran friend of mine wouldn't even listen to me discuss Slaughterhouse Five with him until I told him about the author's background (apparently during 'Nam Vonnegut was considered a traitor by draftees?).
All these life experiences — who actually succeeds when their only goal is to become a writer?! Empty words, empty people.
Do you have a better Gene Wolfe introductory recommendation? His 2nd book for me to read?
Wolfe is one of my favourite writers, and I highly recommend New Sun.
That said, it is a challenge for a lot of readers. It's a single, very long novel that introduces a complicated and mysterious cosmology that is rarely fully understood until the second or even third reading. A common saying among fans is that you don't read Wolfe, you re-read him. It requires a certain amount of patience.
An easier intro is The Fifth Head of Cerberus, which is also one of my favourite novels. It's very short, but its puzzle box structure is no less satisfying or challenging than New Sun. Parts of the fun is figuring exact out who is narrating the three different stories that make up the book — it's probably not who they claim to be — and exactly what happened.
Thinking about the article's reference to Herman Melville, Ursula LeGuin actually called Wolfe "our Melville", "our" meaning science fiction writers as a group.
I don't know what you like, but in addition to the Book of the New Sun I really enjoyed his three-volume "Soldier" series, about a mercenary in ancient Greece who suffers from Memento-style amnesia (although this series far predates Memento).
The "Wizard Knight" series (two books) I also really enjoyed if you like something with a bit more fantasy bent.
I’m one of those, although I tend to de-emphasize this fact in most of my social interactions (not just writing related) since people tend to react weirdly to learning that I program computers for a living. Instead, the bio for my writings says, “[he] spends his days as an insignificant cog in the machinery of corporate America.”
I’ve had modest success with a few dozen stories and poems published over the last dozen years, but I don’t expect it will ever be a major source of income for me.
McCarthy was famously impoverished for most of his life. He apparently spent most of his money buying books. Late in life, the movie income from No Country for Old Men and The Road made him a multimillionaire, and his spending was apparently quite wild from then on, buying endless amounts of cowboy boots and tweed coats, as well as a large collection of vintage cars. [1]
His book Suttree is effectively biographical, down to his impoverished 1950s Knoxville streetslang.
I haven't read anything else, but the film No Country for Old Men is incredible storytelling (and I only started Suttree after a /hn/article from a few months ago described the polymathic dismantling of his impressive library).
His short article The Kekulé Problem sheds serious insights (to me, at least) on whether or not LLMs can, alone with language, ever become truly conscious (are words, alone, enough?). Not the main point of the article (rather: about lucid thinking/states leading to wordless solutions presenting themselves to "discoverers," dreaming).
I tried gpui recently and I found it to be very, very immature. Turns out even things like input components aren't in gpui, so if you want to display a dialog box with some text fields, you have to write it from scratch, including cursor, selection, clipboard etc. — Zed has all of that, but it's in their own internal crates.
Do you know how well gpui-component supports typical use cases like that? Edit boxes, buttons, scroll views, tables, checkbox/radio buttons, context menus, consistent native selection and clipboard support, etc. are table stakes for desktop apps.
I do think gpui needs a native input element (enough that I wrote one (https://github.com/zed-industries/zed/pull/43576) just before they stopped reviewing gpui prs) but outside of that I think it is pretty ok and cool that gpui just exports the tools to make whatever components you need.
I could see more components being shipped first party if the community took over gpui, or for some crazy reason a team was funded to develop gpui full time, but developing baseline components is an immense amount of work, both to create an maintain.
Buttons (any div can be a button), clipboard, scroll views (div, list, uniform_list) should all already be in gpui.
Echopraxia is great. I never understood those who thought it was disappointing. Blindsight is wonderful, but Echopraxia is possibly the more inventive one. It certainly pulls the narrative in a different direction.
I also really, really recommend The Freeze-Frame Revolution. It's about the crew on an starship trying to stop the rogue (sort of) AI that runs everything, the twist being that the crew is constantly under surveillance and must periodically hibernate in shifts for months or years at a time. It's a novella plus a handful of short stories set before and after the novel (all available for free on Peter Watts' website). Be warned, it's one bleak, dark universe.
Also, don't miss out on "The Colonel" (also on his website), a standalone short story that also happens to be a direct sequel to Blindsight.
Solaris by Lem is perhaps the one above all. Lem wrote several of these "inscrutable alien first contact" novels: His Master's Voice, The Invincible, Fiasco, and Eden are basically all variations on this theme, each one unique and highlighting a different aspect of humans' inability to understand the universe. The last three are a little dated now, but still enjoyable to read. HMV is rather dry, a Borgesian essay on an investigation into an alien signal, with lots of references to fictional scientific papers. (Len also wrote two collections of very Borgesian essays that are basically reviews of fictional books: A Perfect Vacuum and Imaginary Magnitude. They're interesting and funny, but I wouldn't put them among his most entertaining work.)
Roadside Picnic by Arkady and Boris Strugatsky is also a masterpiece. They also have a few stories about unseen aliens manipulating the history of humanity by placing traps or transforming humans into infiltrators. The Max Kammerer books (e.g. Beetle in the Anthill) involve this storyline and are very good, probably not well known today.
I tried Tchaikovsky (both Children of Time and Shroud) and found him to be completely unengaging as a writer. Just really dull writing and flat characters. Watts and Reynolds are much better writers. Watts in particular can really pack a punch.
reply