Skip to content - My tmux test SIGKILLed the agent running it3 min read
- My agent claimed a test suite that did not exist4 min read
- Deleting a clean git worktree broke a live agent session's shim4 min read
- The git check that proves isolation isn't the one you'd reach for4 min read
- Park the decision, not the task5 min read
- My alarm was an `if` statement, and my backend wasn't in it5 min read
- When you're straining to recall the command, that's the bug3 min read
- Merged is not deployed2 min read
- You can't un-queue a poller2 min read
- The tmux window title lied to me3 min read
- The merge gate was built to avoid loops, not to define done3 min read
- The -A flag that unhung my launchd agent2 min read
- Small request 200, big request 429, same account, same second3 min read
- Outdated means the lines moved, not that you fixed it3 min read
- Half my agents never got the memo3 min read
- The glob that ate the rest of my file3 min read
- The comment count was lying to me2 min read
- systemctl show lied to me about my own env var2 min read
- The outdated comment that wouldn't die2 min read
- An old timestamp is not a dead backup3 min read
- The coordinator shouldn't be running grep3 min read
- My rescue daemon said 'LIVE' while seeing nothing2 min read
- My fleet boss reported 22 stuck panes instead of fixing them3 min read
- The crash loop that passed every health check3 min read
- Zero events, exit 1, one second: the failure is startup, not your code3 min read
- When a quick fix becomes a dig, send someone else down the hole3 min read
- My agent kept SSHing into the box it was already running on2 min read
- Ask the pane you're in, not the one tmux is looking at2 min read
- My rescue script typed a command into Claude's chat3 min read
- Merged is not running2 min read
- My agent burned the SSH lockout budget guessing keys2 min read
- The zombie PR loop: why my agents kept working after the job was done3 min read
- The tmux title said 'Debug QUIC error.' It was three days out of date.3 min read
- My file sync committed a delete of 803,100 files. Then it tried to push it everywhere.8 min read
- Name the pane, not the UUID3 min read
- The admin override is a different trust context3 min read
- Empty `which` means 'not in PATH', not 'not installed'2 min read
- The process I killed was alive. It just had a different name.3 min read
- The pipeline is the review3 min read
- 'Go all the way' does not mean merge3 min read
- scutil --dns lied to me2 min read
- Don't send your agents on a scavenger hunt3 min read
- The rescuer couldn't see the lifeline3 min read
- My resume hook came back alive and froze on the first question3 min read
- My agent said it was blocked. It had the keys the whole time.3 min read
- Never run git checkout in a loop's working directory3 min read
- The loop is not allowed to decide it's done3 min read
- Green CI is necessary, not sufficient3 min read
- My fleet boss asked permission for a chore2 min read
- A credential clobber looks exactly like a rate limit3 min read
- My agent talked to a wall for 55 ticks3 min read
- Green didn't mean seeing: the night my rescue daemons watched nothing3 min read
- 'Threads resolved' is not 'the fix is in the code'3 min read
- Make your AI reviewer argue with itself3 min read
- CODY has to prove itself wrong first3 min read
- My tmux-resurrect snapshot lied, so I rebuilt the Claude fleet from jsonl mtimes4 min read
- My macOS agent workers went dark until I moved them from LaunchAgent to LaunchDaemon4 min read
- Finish, don't stage: what I want from an agent on the night shift4 min read
- Green on mocks is not done6 min read
- Visibility is not theater4 min read
- Git worktrees ate my edits, so we switched to dedicated machines for agent isolation6 min read
- Skills are just the beginning: the four-layer agent stack6 min read
- Your CLAUDE.md is probably making your agent worse5 min read
- Building an AI analysis agent in hours: a no-code approach with Lovable and n8n8 min read