> The name is borrowed from the Wallfacers of Liu Cixin's The Dark Forest — people entrusted with plans too sprawling for anyone else to follow.
Which misunderstands the book. The point of Wallfacers is that they cannot share their plan with anyone because the enemy is spying on them at all times.
There’s Rust/Tantivy-based search and a whole host of utilities to view, resume, trim (to clean up context when close to limit), port to Claude/codex, move accounts, etc.
The search has a json mode for agents, and a corresponding skill so you can just ask Claude/codex to search for the “latest session where I worked on xyz”.
I work in a massive monorepo with a bunch of different services. Over the last month, I realized I was starting Claude Code sessions everywhere. I would hit a weird issue, remember I had already solved it with the agent a few days ago, and then realize I had no way to easily find that specific session.
CC stores these as JSON files, keyed by the directory you happened to be in at the time.
I built Wallfacer because I needed a way to search and resume these sessions without messing with the source of truth. It operates as a strictly read only overlay. It just reads the local files, extracts the working directory and first prompt, and lets you manage the metadata in a local SQLite database.
I have a codebase that is so large that after of couple of weeks this and after working on similar stuff all over the place this becomes impossible, and expensive.
Nice, I will give this a go. This is what /resume should be but because sessions are tied to directories it's very easy to lose track of where you started what. Until now I've been asking Claude 'what was the conversation we had about [X]' which works but is clunky in the extreme.
Same thing bit me. What fixed it was keeping the session id next to the checkout instead of letting the directory be the key. Small local db, one row per worktree. Resume then doesn't care where I'm standing, it just continues that transcript. I've got worktrees I haven't touched in weeks that still pick up fine.
One thing I didn't expect: forking a conversation copies the history into a new transcript. So anything you count per transcript, cost in my case, double counts that prefix. Resuming the same id doesn't.
I see the need for such a tool. I just ask myself: why should I use a brand-new vibe-coded tool if I can just vibe-code my own? One that's fine-tuned exactly for my use case?
That's not a rant against Wallfacer. It's just an observation about the state of the open-source world in 2026...
In a world where code is cheap, code no longer equals effort. These days, I see consistency, attention to detail, long term support, and community engagement as the signals to read. Not to say this project will never become that, as it may! However until I see those signals as a clear sign of momentum, it's just noise.
I'll never vibe-code PostgreSQL. I don't want to maintain my own Kubernetes codebase. I happily pay for GitLab Enterprise and Authentik Enterprise.
That said, these small end-user console apps are just noise.
It's so cheap to build something that fits your exact use case that almost everyone does it instead of contributing to an existing project or reading through their AI generated documentation...
And posting it on Hacker News feels like the software developer equivalent of an influencer posting a picture of their lunch.
This is exactly what I did. I create tooling to make it easier for me to manage multiple multi-project swarms across multiple clis (claude, codex, opencode), with queuing, and team sharing.
It was.. surprisingly a little bit under a day's work. It's now heavily tied into our way of working and thinking. This is the upside of the current world.
What used to be time spent in personal, throwaway(ish) shell scripting has now become the same, but for more full fledged tooling. It doesn't need to be consumable by others - but it does need to make my/our life/ves better
this a cynical interpretation of it. It reduces OSS to free source code, not as in free speech, but as free beer.
OSS was always about collaborative engineering process. Together we bet the system. Hacking for freedom. Now we reduced it to the modern individualism.
I didn’t quite understand. I use Tmux extensively and have tons of sessions running different panes etc. But how does Tmux solve the session search problem, maybe I’m not aware of some Tmux feature?
The name is pretty ominous if you've read the Three Body Problem series. I don't want my agents planning against me in secrecy.
Just out of curiosity, why did you pick that name? Is there anything wallfacey about this project I'm not getting?
From the README:
> The name is borrowed from the Wallfacers of Liu Cixin's The Dark Forest — people entrusted with plans too sprawling for anyone else to follow.
Which misunderstands the book. The point of Wallfacers is that they cannot share their plan with anyone because the enemy is spying on them at all times.
maybe it is deliberate to throw off the spies.
Lol, I understand. I was reading the book and I was at the part where the wallfacer project was announced. I did not want to think too much more :D
Related: several months ago I made a TUI session search/manager tool as part of my suite of Claude/codex productivity tools:
https://pchalasani.github.io/claude-code-tools/tools/aichat/...
There’s Rust/Tantivy-based search and a whole host of utilities to view, resume, trim (to clean up context when close to limit), port to Claude/codex, move accounts, etc.
The search has a json mode for agents, and a corresponding skill so you can just ask Claude/codex to search for the “latest session where I worked on xyz”.
I work in a massive monorepo with a bunch of different services. Over the last month, I realized I was starting Claude Code sessions everywhere. I would hit a weird issue, remember I had already solved it with the agent a few days ago, and then realize I had no way to easily find that specific session. CC stores these as JSON files, keyed by the directory you happened to be in at the time.
I built Wallfacer because I needed a way to search and resume these sessions without messing with the source of truth. It operates as a strictly read only overlay. It just reads the local files, extracts the working directory and first prompt, and lets you manage the metadata in a local SQLite database.
I just tell claude "Few days ago we worked on xxx, find the session and let's resume."
I have a codebase that is so large that after of couple of weeks this and after working on similar stuff all over the place this becomes impossible, and expensive.
Nice, I will give this a go. This is what /resume should be but because sessions are tied to directories it's very easy to lose track of where you started what. Until now I've been asking Claude 'what was the conversation we had about [X]' which works but is clunky in the extreme.
Same thing bit me. What fixed it was keeping the session id next to the checkout instead of letting the directory be the key. Small local db, one row per worktree. Resume then doesn't care where I'm standing, it just continues that transcript. I've got worktrees I haven't touched in weeks that still pick up fine.
One thing I didn't expect: forking a conversation copies the history into a new transcript. So anything you count per transcript, cost in my case, double counts that prefix. Resuming the same id doesn't.
you can name Claude sessions using /rename. also tmux is your friend.
reminds me of good old days of opensource, when we use to search everything on github, but good try, keep building
I see the need for such a tool. I just ask myself: why should I use a brand-new vibe-coded tool if I can just vibe-code my own? One that's fine-tuned exactly for my use case?
That's not a rant against Wallfacer. It's just an observation about the state of the open-source world in 2026...
In a world where code is cheap, code no longer equals effort. These days, I see consistency, attention to detail, long term support, and community engagement as the signals to read. Not to say this project will never become that, as it may! However until I see those signals as a clear sign of momentum, it's just noise.
Yes, absolutely.
I'll never vibe-code PostgreSQL. I don't want to maintain my own Kubernetes codebase. I happily pay for GitLab Enterprise and Authentik Enterprise.
That said, these small end-user console apps are just noise.
It's so cheap to build something that fits your exact use case that almost everyone does it instead of contributing to an existing project or reading through their AI generated documentation...
And posting it on Hacker News feels like the software developer equivalent of an influencer posting a picture of their lunch.
This is exactly what I did. I create tooling to make it easier for me to manage multiple multi-project swarms across multiple clis (claude, codex, opencode), with queuing, and team sharing.
It was.. surprisingly a little bit under a day's work. It's now heavily tied into our way of working and thinking. This is the upside of the current world.
What used to be time spent in personal, throwaway(ish) shell scripting has now become the same, but for more full fledged tooling. It doesn't need to be consumable by others - but it does need to make my/our life/ves better
isn't this the open source dream? Fork it and make it your own
this a cynical interpretation of it. It reduces OSS to free source code, not as in free speech, but as free beer.
OSS was always about collaborative engineering process. Together we bet the system. Hacking for freedom. Now we reduced it to the modern individualism.
Why do people keep trying to reinvent tmux?
I didn’t quite understand. I use Tmux extensively and have tons of sessions running different panes etc. But how does Tmux solve the session search problem, maybe I’m not aware of some Tmux feature?
Try the left arrow in Claude Code and now you have session search. Or claude --resume, down arrow, then Ctrl-A to see everything
I know about these. My question was about Tmux. Also the Claude resume search is not full text, it only matches session names.