Use when changing subprocess execution, cancellation, process-tree cleanup, orphan reaping, or Windows child hardening.
65%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Passed
No findings from the security scan
Fix and improve this skill with Tessl
tessl review fix ./.agents/skills/process-lifecycle/SKILL.mdContext, Concurrency, and Processes
context.Context through long-running, subprocess, and networked work; prefer exec.CommandContext; use derived contexts and timeouts for cleanup and HTTP calls.shellenv lifecycle configurator. Agents and other subprocesses use ConfigureShellCommand: it creates a process-tree boundary and installs cmd.Cancel to kill the whole tree, so grandchildren (test workers, build watchers) cannot outlive cancellation and hold the next run's worktree locked. Repository commands.* and custom gates use ConfigureCooperativeShellCommand: on Windows it sends targeted CTRL_BREAK, preserves up to a three-second cleanup window, then retains the same forced Job Object fallback. Regressions: TestConfigureCooperativeShellCommandCancelAllowsCleanup, TestConfigureCooperativeShellCommandCancelForcesNonCooperatingTree.cmd.Cancel covers only cancellation; on clean exit or error the group is not reaped, and leaked grandchildren accumulate until the OS OOM-kills the daemon (surfacing as daemon crashed during execution with no stack trace). Use shellenv.RunShellCommand / OutputShellCommand / CombinedOutputShellCommand for one-shot commands, or StartShellCommand plus TerminateShellCommandGroup when handling pipes manually; the helper doc comments in internal/shellenv own the details. ConfigureShellCommand also installs a 5s cmd.WaitDelay backstop so a grandchild holding an inherited pipe cannot wedge cmd.Wait forever. For Claude, Codex, and Pi, a cleanly exited agent's completed result remains usable when that backstop closes an inherited pipe; without a completed result, the parse failure is retained and reports whether a result event arrived. Claude also stops a process still open claudeResultGrace after its completed result (background work holds print mode open); a RequireCompletedChildren turn fails there and never counts that result. Regressions: TestClaudeAgent_BackgroundWorkAfterResult, TestNativeAgents_FinishedResultSurvivesEscapedStdoutHolder, TestNativeAgents_UnfinishedRunKeepsParseErrorOverWaitDelay, TestCodexEmptyResultEventIsReportedAsReceived, TestCodexAgent_Run_ReapsLeakedGrandchildOnCleanExit, TestRunShellCommandWithEnv_ReapsGrandchildOnCleanExit, TestTerminateShellCommandGroup_*.setsid(2)/setpgid(2) (agent CLIs sandboxing their tool runners, any daemonizing worker script) leaves the group, and after its parent exits nothing lineage-based can name it again - it burns CPU and holds a deleted worktree's cwd forever. internal/procreap is the identity-based backstop: it matches a process by the run worktree its cwd resolves under (deliberately never argv, which a legitimate git worktree remove also carries), never touches pid<=1/itself/its ancestors, spares worktrees whose run is still pending or running, and escalates SIGTERM to SIGKILL only after a grace period. Reach is <NM_HOME>/worktrees by path shape plus exactly the run worktrees a caller names from run records (Options.Worktrees), never a configured worktree root by shape - an operator's own directory is unmatchable unless a run row names it. Every site that removes a run worktree sweeps it first through procreap.SweepRunWorktree(s) (run cleanup and setup failure via RunManager.removeRunWorktree, startup cleanup, eject), scoped and therefore without age floor or run-active check; the unscoped startup sweep in recoverOnStartup keeps the orphanProcessMinAge floor. All best effort. Windows needs none of this - job objects contain the whole tree - so the platform layer reports an empty table. Regressions: internal/procreap, TestSweepOrphanRunProcessesReapsFinishedRunAndSparesActiveOne, TestSweepRunWorktreeProcessesReapsLeakedChildAtRunCleanup, TestTerminateShellCommandGroup_AsksBeforeKilling, TestTerminateShellCommandGroup_EscalatesWhenSIGTERMIsIgnored.no-mistakes.exe daemon run --root <root> directly. Its XML definition sets DisallowStartIfOnBatteries and StopIfGoingOnBatteries to false and ExecutionTimeLimit to PT0S, because Task Scheduler's short-job defaults otherwise stop a long-running daemon on battery and after a fixed runtime; keep those settings. The binary embeds an application manifest with consoleAllocationPolicy=detached, so Windows 11 24H2 / Server 2025 (build 26100) and later allocate no console when there is none to inherit, while a terminal, hook, or agent still shares its console. The manifest is owned by internal/winres, which renders cmd/no-mistakes/rsrc_windows_{amd64,arm64}.syso (make winres; make lint and a go test fail on drift); never commit an object from another tool. Older builds ignore the manifest, so daemon.PrepareRunStdio, the first thing daemon run does (and only daemon run, never another entry point such as the log-sink or a CLI sharing a hook's or agent's console), frees a console only this process is attached to (a terminal's shared console stays) and points stdout/stderr without a usable handle (NULL under the manifest start, or the released console) at the bootstrap log, as the detached start path does. The policy also overrides CREATE_NEW_CONSOLE, so the cooperative-command helper (the same binary) allocates its private console itself, without a window, before starting its target (ensureWindowsCooperativeConsole), or CTRL_BREAK never reaches the target. Keep the task's process the daemon itself: a launcher or conhost --headless wrapper breaks schtasks /End, inspectWindowsManagedDaemon, and the argv scans in collision.go. Regressions: internal/winres (TestObjectCarriesOnlyTheManifest, TestCommittedObjectsMatchGenerator, TestNoMistakesExecutableCarriesExactlyTheManifest, TestNoMistakesGetsNoConsoleFromAConsolelessParent), TestConfigureCooperativeShellCommandCancelAllowsCleanupWithShippedHelper, TestReleaseSoleConsoleFreesOnlyAConsoleOfItsOwn, TestPrepareRunStdio*, TestRunPreparesDaemonStdioBeforeTheDaemonRuns, TestOnlyDaemonRunPreparesDaemonStdio, TestRenderWindowsTaskXMLScopesLogonTriggerToInstallingUser, TestRenderWindowsTaskXMLAllowsLongRunningOnBattery, TestInstallWindowsTaskRegistersUserScopedLogonTask.Start-Process without -NoNewWindow, a double-click) - gives every console child a window of its own, so route every console child through winproc.Harden(cmd) (no-op elsewhere, idempotent, preserves existing creation flags) or a console window flashes per child (#287). shellenv.ConfigureShellCommand already calls it; one-shot commands built directly must call it themselves. Regressions: TestHarden* in internal/winproc.kunchenguid/no-mistakes
3d2d4bf
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.