frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Show HN: Ctxwitch – Git tells you what changed; this tells you what it'll do

https://github.com/ctxwitch/ctxwitch
1•kulkarniamey•56m ago

Comments

kulkarniamey•36m ago
Author here. I left in all those shipping changes to the agent - a prompt modification, a more lenient guardrail, some increased temperature -that altered the behavior of the agent without altering its execution, which flew past code reviews and CI and only came to light in prod. ctxwitch diffs changes made to the prompt/model/tools/RAG/memory/guardrails of an agent and categorizes their behavioral effect in 12 different ways at 5 levels (cosmetic -> breaking). A removal of guardrail or reversal of any rule gets marked as Breaking, and CI system gets notified that a more rigorous evaluation is needed, while cosmetic modifications sail through. It works both locally and on Github actions, takes about 100ms, deterministic (no agent executions, traces, or LLMs are used; there's an option to use LLM judges for truly subjective cases).

But in an attempt to address the obvious problem: a configuration difference doesn’t prove something was actually run, but rather that there was intent to make it happen. Therefore, I designed it as a triage/gate - "this is something worth looking at".

pip install ctxwitch witch tour #disposable sandbox, no api

Apache 2.0. I really would like to know for those running ai agents in production what diffs have caught up with you that something a diff like this should have spotted, and what more would it need to pickup to get into your CI?

I built my own filing cabinet

https://www.alexstephen.me:443/writing/filing-cabinet/
1•rambleraptor•47s ago•0 comments

Building Xbox games with Three.js [video]

https://www.youtube.com/watch?v=ETfSiUHehY4
1•arbayi•1m ago•0 comments

NHTSA orders Tesla to prove its Cybercab is legal to sell, under oath

https://electrek.co/2026/09/15/nhtsa-tesla-cybercab-special-order-fmvss-certification/
1•Tomte•4m ago•0 comments

Rune raises $40M to deploy off-grid compute capacity at renewable energy sites

https://fastcompany.com/91607727/what-if-ai-data-centers-didnt-need-new-power-plants
1•utiiiD•4m ago•0 comments

Git Worktree Gotchas

https://www.olafalders.com/2026/09/16/git-worktree-gotchas/
1•oalders•4m ago•0 comments

Show HN: Free WhatsApp MCP (+UI) – Give Your AI Agents Access to WhatsApp

2•fabian_shipamax•6m ago•0 comments

Django: Simple Streaming / SSE with Mercure

https://blog.tmk.name/2026/09/16/django-simple-streaming-sse-with-mercure/
1•tristanmk•6m ago•0 comments

The smallest possible Linux distribution

https://distrowatch.com/weekly.php?issue=20260914
1•utiiiD•7m ago•0 comments

Show HN: Agentbox – Teleport your repo into sandboxes with no worktree juggling

https://github.com/madarco/agentbox
2•madarco•8m ago•0 comments

Show HN: Gokudo Wiki – a database and tools site for a new Roblox game

https://gokudowiki.com/
1•richardharmer•9m ago•0 comments

LLM based CI pipeline code generator for DSCI

1•melezhik•10m ago•0 comments

Show HN: Wenlan – a living wiki AI keeps current without overwriting your edits

https://github.com/7xuanlu/wenlan
1•h164654156465•11m ago•0 comments

Mayfly Chat: Transient Chat for Agents

https://blog.exe.dev/mayfly-chat
1•indigodaddy•13m ago•0 comments

Show HN: Plutus – click a release, see what moved on your cloud bill

https://demo.plutus-cloud.com
1•thechenderson•13m ago•0 comments

Estimating the Cost of Combat Operations Against Iran [pdf]

https://www.cbo.gov/system/files/2026-09/62756-Iran.pdf
2•amarcheschi•13m ago•0 comments

Banana Mode

https://hatchet.run/bananas
1•noleary•16m ago•1 comments

There is a channel to 900M weekly users. What goes in it?

https://www.lesswrong.com/posts/gJJ9YHzuBvwAXrthW/there-is-a-channel-to-900m-weekly-users-what-go...
1•ddp26•17m ago•0 comments

Archive Builder

https://apps.microsoft.com/detail/9n5hjq99k6wq?hl=en-US&gl=US
1•RENiXTech•18m ago•0 comments

Self-evolving agents need pain, reflection, and sleep

https://medium.com/@robertindie2016/self-evolving-agents-need-pain-reflection-and-sleep-2c63e6bd6630
1•aaronrobert•19m ago•0 comments

Ask HN: Has anyone measured how often agents use the skills you ship?

1•sohaibtariq•19m ago•0 comments

Mojo compiler is open for open source contributions [Mojo]

https://forum.modular.com/t/mojo-compiler-is-open-for-open-source-contributions/3499
1•ivell•20m ago•0 comments

Predictive database benchmarks vs. RF, AutoML, Elastic etc., up to 10M scale

https://aito.ai/docs/api/v2/benchmarks/
1•arauhala•21m ago•0 comments

TypeScript team chose Go over Rust

https://www.thetrueengineer.com/p/typescript-team-chose-go-over-rust
1•adletbalzhanov•21m ago•0 comments

U.S. has deployed space-control weapons in orbit, Air Force secretary says

https://spacenews.com/u-s-has-deployed-space-control-weapons-in-orbit-air-force-secretary-says/
2•alakra•21m ago•0 comments

Show HN: TabHop I just wanted Command+Shift+Up/Down to switch tabs

https://chromewebstore.google.com/detail/tabhop/jhcabemalinecggplldpoikighephkjn
2•jackwalkerlabs•22m ago•1 comments

Show HN: ctx – Git blame that returns the original agent transcript

https://ctx.rs/pro/
2•luca-ctx•23m ago•0 comments

Pangram: The Most Reliable AI Detector Has a Human Problem

https://www.bloomberg.com/news/features/2026-09-16/pangram-ai-detection-tool-tries-to-prove-tech-...
1•pretext•24m ago•0 comments

Meta to report child abuse material directly to Indian authorities

https://www.bbc.com/news/articles/cqx2zpn2y97eo
1•vinni2•27m ago•0 comments

In Germany's east, old ties to Moscow fuel AfD rise

https://www.ft.com/content/0aa2e6af-195d-40cc-a813-3e130cbb5553
3•alephnerd•27m ago•0 comments

Insights on AI from the Reformation

https://www.amytis.io/blog/insights-on-ai-from-the-reformation
1•JohnHammersley•28m ago•0 comments