... Switching between clients could reproduce that combination. The screen model now uses avt, the terminal model used by asciinema. If a parser fault occurs, the screen can be rebuilt at the same grid size and a poisoned mutex recovered. Scrollback may be lost for that terminal, but the failure should no longer make the live terminal permanently inaccessible through this path. The release also fixes output being held up by an unanswered redraw, and moves screen snapshot work, output encoding ...

... moment in every long agent session that almost nobody notices: compaction. When the conversation approaches the context limit, the harness has a model summarize everything so far and swaps the summary in for the real thing. Your scrollback still shows the full conversation. The model's world is now the summary. This explains one of the most confusing agent behaviors: hours into a session, it stops honoring a decision you both settled early. You scroll up and the agreement ...

... week three, every "small change" was touching six files I didn't expect, and the agent kept re-deciding architecture choices we'd already settled two days earlier - because the only record of those decisions was buried in scrollback. What actually fixed it for me wasn't a better prompt or a bigger context window. It was writing a spec before touching the agent at all - a short doc covering what the feature needs to do, the inputs/outputs, the constraints ...

... quietly, the most accurate record any of us keeps. Nobody games their own calendar. It just sits there being right. 3. The timesheet / invoice. Filled in Friday at 6pm or worse, at month-end from memory, from Slack scrollback, from vibes. Then it becomes the number a client actually pays. So, three questions for anyone running or working in a small team: Do your daily updates still earn their slot, or have they become theatre? Has anyone actually made ...

... from Moss Ab Mirande-Ney(@moss_ab_mirande_ney): @thomas_shellby1  You are right about the real cost, and it took us a while to see it. The agent stopping is annoying, coming back to a 4000 line scrollback and not knowing what happened is what actually wastes the afternoon. tmux on a VPS is exactly where we started and it holds fine for one person. Two things broke it for us. A second person attaching to the same ...

... feature slower. rate limits are the only mechanism that reliably catches the 'ill just trust it and keep going' pattern. worth using them as the checkpoint instead of routing around them. Comment from Raffay Sajjad(@raffay_sajjad): The decision vs reasoning split matches something I do without ever having a name for it. Before I stop for the day I write one line per thing I changed and why, not what, git already has that part. What breaks ... ... below with a memory folder, worth combining the two. Write the note the way you already do, but drop it into a file the next session reads on start instead of carrying it in your head or a chat scrollback

... left reconstructing what actually happened, Blocks is my answer https://www.verlox.app . Every command you run becomes its own card with the command, its output, and a quiet line that tells you the outcome in plain English. No digging through scrollback to figure out whether something worked. A few things I'm proud of in this release: Fix this. When a command fails, one button sends that exact run, the command plus its own output, to the Al and it gets ...

... actual insight, and it's bigger than it looks: making the notification say why it's waiting is what turns a ping into a triage. "Done" makes you tab back to find out what "done" means. "Needs approval vs finished vs

... just shipped xshell a native desktop app that puts every Claude Code session, project, and cost in one window. One click to resume any past session with full scrollback. Built with Tauri 2 + Rust. Works on Windows, macOS, Linux. Honest feedback (especially the harsh kind) really helps https://xshell.sh https://github.com/MertPROJ/xshell ...

... terminal rendering loop suspends when a tab loses focus, so there's zero GPU/paint overhead. Each tab holds: 1 persistent SSH connection a lightweight open socket, near-zero CPU when idle 1 terminal buffer ~2-5 MB depending on scrollback history So 10 open sessions 10 sockets + ~20-50 MB of memory. Comparable to running multiple tabs in any native terminal emulator. SSH connections are event-driven, so inactive sessions consume virtually no CPU they only wake on incoming data ...