{"repo":"NousResearch/hermes-agent","public_id":"112344","display_label":"Desktop backend respawn after stale exit failures","summary":"These issues and fixes address a bug where the desktop app's local backend fails to respawn after being classified as a stale exit, leaving the UI running without an engine. The work focuses on making backend recovery logic more robust and independent of pool state.","subsystem":null,"counts":{"members":5,"open_issues":0,"open_issues_distinct":0,"open_prs":0,"closed_issues":1,"merged_prs":1,"attempt_prs":3},"landed_fix":{"status":"merged_to_main","basis":"structural","pr":113180},"resolution":{"state":"resolved","by":"pr-113180"},"freshness":{"state":"verified","last_verified_at":"2026-09-17T00:19:07+00:00","is_stale":false},"grouping":{"backbone":"STRUCTURAL_SPANNING","link_density":0.4},"fix_edges":[{"issue_number":112344,"pr_number":113180,"tier":"verified","basis":"closing_ref","confidence":0.95,"recoverable_from_timeline":true}],"members":[{"kind":"issue","number":112344,"title":"[Bug][Desktop/Windows] Desktop app logs \"Ignoring stale Hermes backend exit (1)\" and never respawns the backend — UI keeps running with no engine (9 h observed)","url":null,"state":"closed"},{"kind":"pr","number":112345,"title":"fix(desktop): recover after backend exit","url":null,"state":"closed","close_reason":{"refs":[{"kind":"pr","value":"#113180","number":113180},{"kind":"pr","value":"#112344","number":112344}],"category":"superseded_by_pr","implication":"superseded"},"merged":false},{"kind":"pr","number":112356,"title":"fix(desktop): respawn when stale backend exit leaves no owner","url":null,"state":"closed","close_reason":{"refs":[{"kind":"commit","value":"8e5b0e2a1b","number":0},{"kind":"pr","value":"#112344","number":112344},{"kind":"pr","value":"#113180","number":113180},{"kind":"pr","value":"#112349","number":112349},{"kind":"pr","value":"#112356","number":112356}],"category":"implemented_elsewhere","implication":"superseded"},"merged":false},{"kind":"pr","number":112989,"title":"fix(desktop): keep primary recovery independent from pool state","url":null,"state":"closed","close_reason":{"category":"superseded_by_pr","implication":"superseded"},"merged":false},{"kind":"pr","number":113180,"title":"fix(desktop): local backend respawns after a stale-classified exit and exit lines carry the backend's last output (#112344, salvage #112349, #112356)","url":null,"state":"merged","close_reason":{"category":"merged","implication":"superseded"},"merged":true,"is_fix_pr":true,"fix_tier":"verified","files_changed":["apps/desktop/electron/backend-claim.test.ts","apps/desktop/electron/backend-claim.ts","apps/desktop/electron/backend-exit-recovery.test.ts","apps/desktop/electron/backend-exit-recovery.ts","apps/desktop/electron/main.ts","website/docs/user-guide/desktop.md"]}],"trust":{"default_tier":"verified","includes_suggested":false}}