Discussion

unsubtlecoder: Interesting, one challenge with other ADEs (nice term btw) like Conductor is that code navigation is terrible and too much emphasis is on a GUI for Claude.We really need the best of both worlds: IDE (powerful like Intellij) + ADE (multitasking code)And how does it compare to other tools like Conductor?

mattv8: Very nice. Does this support GitHub Copilot subscriptions (oauth/hmac) or do you have plans for it? That would make or break for me because of the API costs.Similarly I built a self-host able replit-like server with RAG but it's more end-user focused than developer focused...

luca-ctx: Yes. The important distinction is that ctx sits above the agent harness, not above the raw model APIs.With something like Cursor, you can use models from OpenAI, Anthropic, etc, but they still run through Cursor’s own agent harness.With ctx, you bring the existing harness itself — Copilot, Claude Code, Codex, and so on — and it keeps its own auth/billing/session model. ctx is the layer around that: worktrees, review, runtime boundaries, merge queue, etc.

Bnjoroge: Agree with this. I find myself switching panes between conductor or codex and zed because of code navigation. Maybe that's what cursor is trying to do in their new version, but I havent tried it yet

bloppe: I don't understand why so many people building agents feel the need to fork and maintain a whole IDE as well

leetvibecoder: Does this solve indexing of codebases like Cursor does, or do you still need tools / plugins like Lumen (https://github.com/ory/lumen) for that in order to work in larger codebases without wasting tens of thousands of tokens on tool calls and brute force guessing with grep?

hmokiguess: how does lumen differ from serena? https://github.com/oraios/serena curious about it seems promising

leetvibecoder: Serena is more about text editing and better code search tooling while lumen is (a) chunking code with tree sitter and (b) storing embeddings (vectors) generated by an embedding model (ideally ones which are for code) which you then can search against. Effectively it‘s RAG for code made available as an MCP server.This reduces tool calls (and thus saves times and tokens) because instead of „trying“ / „guessing“ names repeatedly, tools like claude code typically get useful search results on the first try.Claude for example may search for „dbal“ via regex, but the function name is „sql“ - semantic search will find that while for regex, claude would try 3 additional guesses before it actually finds what its looking for. Hope this helps!

hmokiguess: Oh thank you for the detailed response, that makes perfect sense, appreciate it and will certainly give it a try now!

Bnjoroge: Looks cool! Two things: I see you mentioned the merge queue, but how exactly do people avoid or resolve merge conflicts when merging work from two or more agents in the separate worktrees? I havent really seen a seamless way to approach this or do people just have the agents work on distinctly unrelated stuff? Secondly, are containers the primary sandboxing appraoch? or do you support vms?

throwif: exactly tmux + claude code + your real editor. the ide tax is real and nobody benchmarks it

SparkyMcUnicorn: Your repo says it's open source, but it's missing the source.https://github.com/ctxrs/ctx

luca-ctx: That is our issues-only repo, similar to Claude Code. It is not open source.

abound: That's a bummer, especially since folks normally use the ".rs" TLD for Rust projects, so the (perhaps accidental) implication from the domain is that this is a Rust project with the source available somewhere.

luca-ctx: It is a rust project, just not source available.

luca-ctx: You can't avoid merge conflicts from happening, but you can definitely empower the agents to self-resolve them.The workflow is like this:1. an agent works in its own worktree2. its changes are green in isolation3. it submits that work to the local merge queue4. the queue replays the change on top of the latest target branch and runs verification5. if it conflicts or fails after replay, the merge is rejected6. the agent can then pull in the new upstream state, resolve the conflict or test failure, and resubmitWe've found that agent-driven conflict resolution via a merge queue works really well in practice. It's almost necessary because of the increase in velocity of changes.Regarding sandboxing approach, containers are primary right now. We do this natively on Linux and with Apple Virtualization Framework (AVF) on Mac. So yes, there is a VM involved on Mac, but it’s not exposed as a separate top-level mode.

Snakes3727: Fundamentally one of my biggest gripes with tools like this is that often you are not working with a single repo in anything beyond simple apps.When I am working with Claude I am often doing it from the root directory of a workspace of dozens of repos. I work with Claude to come up with a plan for implementing a feature and it investigates and plans.That plan often encompasses multiple repositories. Claude then turns large scale plans into smaller issues, or tickets as artifacts.

luca-ctx: We’ve felt the same thing and tried to make ctx work well in multi-repo setups.There are basically two ways to approach it:- If one repo is primary and the others are mostly reference material, use workspace attachments. That lets the agent work in one repo while still being able to read the others. I do this a lot with dependency/source repos. - If the work genuinely spans multiple repos, just initialize the workspace at the parent directory that contains all of them. The harness still sees the same filesystem layout it normally would, so Claude/Codex/etc. can plan and work across repos the same way.The main caveat is that some features are naturally more repo-specific. Merge queue is the obvious example, since landing and replay are much cleaner when there is one target repo/branch model.

iddan: What’s preventing you from putting all of those in a single parent directory and boot into it?

famouspotatoes: Great tool so far- it feels deeply considered.

sspiff: What is the point of hosting a GitHub repo[0] with nothing in it but some links to your domain? There's no code, no license, no nothing.[0] https://github.com/ctxrs/ctx