Amp adds customizable keybindings and Ctrl+S mode switcher in the CLI
Amp shipped customizable keybindings and a mode switcher that describes each mode, and showed a .amp/plugins plus .amp/settings.json workflow for commit-and-archive macros. The release makes thread management and local customization part of the CLI surface instead of a fixed default UI.

TL;DR
- sqs shipped two long-missing CLI controls in Amp: customizable keybindings via
amp config keymapand a Ctrl+S mode switcher that explains each mode. - sqs then used the new keymap support to generate a
commit-and-archivemacro as an Amp plugin, wiring.amp/plugins/to.amp/settings.jsonin about 30 seconds. - According to Amp's manual, Amp now documents both the new keybinding surface and the mode switch flow, including Ctrl+S in the CLI and Alt+D for reasoning-effort changes.
- sqs framed the customization push as a response to people juggling more simultaneous Amp threads, while Agents, Everywhere says the new app and remote control surfaces were built for exactly that workload.
You can jump straight to the manual's keybindings section, inspect the Plugin API, and compare the small UX ship here with Amp's broader rebuild and multi-agent app push. The weirdly useful bit is that the launch post is a tweet demo, not a changelog entry: the clearest workflow example is sqs's plugin macro walkthrough, and the video even hints at unshipped web UI changes via sqs's follow-up.
Keybindings
Amp's new keybinding surface is finally user-editable. sqs pointed users to amp config keymap, while Amp's manual now lists Ctrl+O for the command palette, Ctrl+S for agent-mode switching, and Alt+D for reasoning-effort changes.
That matters because Amp had already been piling more control into keyboard flows. Command Palette, Not Slash Commands moved CLI actions into a persistent palette, and customizable bindings turn that from a fixed interface into one you can actually reshape.
Mode switcher
The other ship is a mode switcher that tells you what each mode does instead of making you memorize Amp's model taxonomy. In Amp's manual, the three modes break down as:
deep: GPT-5.5 with extended reasoning for complex problemssmart: unconstrained state-of-the-art model userush: fast, low-token GPT-5.5 with no reasoning for smaller tasks
The docs previously told users to switch modes through Ctrl+O and a mode command. sqs adds a dedicated Ctrl+S entry point and a descriptive chooser on top.
Plugins
The best demo of the release is not rebinding a key. It is using Amp to write a new command for itself.
sqs showed a prompt that creates .amp/plugins/commit-and-archive.ts, registers a command that sends a "commit the changes, then archive this thread" message, and binds it in .amp/settings.json to <leader> c. After plugins: reload, the macro runs inside the current thread.
That workflow lines up with the official Plugin API: plugins are long-lived JavaScript or TypeScript programs, they live in .amp/plugins/ or ~/.config/amp/plugins/, and they can register commands, tools, UI elements, and event handlers. Amp, Rebuilt described that API as part of the new CLI architecture.
Thread archiving
The macro demo also surfaces where Amp's product direction is going. sqs said users are now juggling "a lot more simultaneous-ish" threads because of remote control and the new app, and the example macro ends by calling amp threads archive on the current thread ID.
That command is now part of the documented workflow. Amp's manual says archived threads disappear from the active list but remain viewable on the web and referenceable by @-mention, while Agents, Everywhere says the web, mobile, and CLI surfaces were rebuilt so users can watch and drive many active agents at once rather than babysit terminal tabs.