Scripts without remembering them
The Scripts tab reads package.json and lists every script. Click Run and Flame opens a dedicated terminal with live output. Saved commands at global, group, or project scope sit in the Commands tab.
Flame IDE's terminal: one-click npm scripts, shell profiles, splits, pinned terminals, pop-out windows, a ports list with kill, and CPU and memory alerts.
A typical day leaves a trail of terminals: a dev server, a test watcher, a database container, a script someone forgot to stop. Flame gives every project its own terminals and then helps you keep track of them: label and pin the long-running ones, split or pop them out, see which process is holding a port, and find the terminal that has been eating a CPU core for ten minutes.
The Scripts tab reads package.json and lists every script. Click Run and Flame opens a dedicated terminal with live output. Saved commands at global, group, or project scope sit in the Commands tab.
Drag a tab onto another to split side by side, rename tabs inline, open a terminal in an editor tab for wide logs, or pop it out into an external window for a second monitor.
Pin a dev server or watcher to keep it one click away from every project. Choose any shell installed on your system from the shell profile menu.
Ports lists every listening port with its process, PID, and owning project. Resources shows CPU and memory for Flame and everything it runs, with alerts for runaway terminals.
EADDRINUSE usually means a forgotten process. The Ports tab lists every listening port on your machine with its address, process, command, PID, and the Flame project that started it.
Filter to loopback or exposed addresses and by protocol, or search by port, process, or command. Pin the ports you check often.
Open a web port in Flame's built-in browser or your default browser, or copy its URL, command, or working directory.
Stop a process gracefully and then by force, stop it with its child processes, or force kill it now. Flame protects its own processes, and suggests stopping the container instead for a Docker-published port.
The Resources tab shows how much CPU and memory Flame and everything it runs is using on Windows, macOS, and Linux. Rows group the Flame window and GPU, the backend, language servers, AI agents, helper processes, and one row per terminal with every process started from it.
Sort by CPU or Memory to find what is keeping the machine busy, and expand a row to see its processes and their command lines. CPU is shown as a percentage of one core, like top or Activity Monitor.
Turn on Always monitor to check every 15 seconds. A terminal tab gets a badge, and a notification links to it, when its processes pass the memory threshold (4 GB by default) or keep about one CPU core busy for longer than the CPU warning time (3 minutes by default).
Usage is checked every 3 seconds only while the tab is open and the window is focused, and not at all otherwise unless Always monitor is on. Both thresholds are configurable in Settings → Terminal.
Terminal output is evidence. Flame lets you hand it to AI Chat without copying and pasting a stack trace.
Use Ask AI or Fix with AI from a terminal to stage a chat with the output. It opens a new chat by default; the arrow menu lets you use the current one. Nothing is sent until you press Send.
Send a terminal's output to the Logs Viewer to parse, filter, and search structured logs instead of scrolling.
When the same command must run in every service, the Broadcast terminal runs it across projects in sequence or in parallel with a status matrix.
Run npm scripts with one click, switch shell profiles, split terminals, pin long-running processes across projects, dock terminals into editor tabs or windows, stop listening ports, and catch terminals that use too much CPU or memory.
Open the Scripts tab and click Run on dev and test. Each starts in its own terminal; rename them so they are easy to find.
Drag the test terminal beside the dev server to split them, pin the dev server, or open a terminal in an editor tab when you need the width.
If a port is already taken, open Ports, find the process holding it, and stop it with its child processes.
Turn on Always monitor. When a watcher gets stuck in a loop, the badge on its tab and the notification take you straight to it.