Refactor scheduler/todo runner onto shared process-runner core

[charlesrhoward]

Summary\n- add a shared main-process runner utility for spawn lifecycle, timeout, force-kill, stdout parsing, and tail capture\n- migrate scheduler task execution to the shared runner while preserving status/error logging semantics\n- migrate todo-runner execution to the shared runner while preserving payload transport and retry behavior\n\n## Validation\n- pnpm run lint:desktop\n- pnpm run typecheck\n- pnpm exec playwright test -c playwright.smoke.config.ts --workers=1\n\nCloses #97

---

Note

Medium Risk
Moderate refactor of main-process child-process lifecycle (spawn, timeouts, kill signaling, and output parsing) for scheduler/todo execution; behavior should be equivalent but subtle timing/error-handling differences could affect job completion and status reporting.

Overview
Introduces a new shared process-runner utility that centralizes child-process spawning, stdin payload writing, stdout line handling, stderr/stdout tail capture, runtime timeouts (SIGTERM then SIGKILL), and force-kill timer management.

Refactors both scheduler and todo-runner to use runManagedProcess plus shared helpers for env-based runtime config and force-kill scheduling, replacing bespoke per-module timers/readline/stdio logic while preserving existing logging/events and error/status derivation (including todo-runner’s payload-transport spawn error handling and retry/pause behavior).

^{Written by Cursor Bugbot for commit e491411. This will update automatically on new commits. Configure here.}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
agent-space-web	Ready	Preview, Comment	Feb 16, 2026 0:22am