abortlisted

# abort — release a persistence loop `implement --from goal|plan` registers a Stop hook that blocks the session from ending while any story is red. This releases it. Pure file operations: no agents, no MCP. ## Run 1. Set `"active": false` in every `<repo>/.nord/state/*-state.json` — the hook treats `!active` as allow-stop. If the user named one mode, only that file; otherwise all. 2. Remove `<repo>/.nord/prd.json`, the story SSOT, so no red story can re-trigger the block. 3. Report which modes were released and how many stories were still red. ```sh for f in .nord/state/*-state.json; do [ -e "$f" ] && python3 -c \ "import json,sys;p=sys.argv[1];d=json.load(open(p));d['active']=False;json.dump(d,open(p,'w'),indent=2)" "$f" done ``` Delete `prd.json` with the file tools, **not with `rm`** — this device's `guard-rm.py` blocks a deletion whose target it cannot resolve against the protected list, and the loop inside a shell variable is exactly that case. ## Verify Re-read one state file and confirm `"active": false`, and confirm `prd.json` is gone. The hook decides on file contents alone; if the write did not land, the session will still refuse to stop and the reason will look like a hook bug. ## Do not delete the state files `active:false` is enough, and nord-hud still reads them for the statusline. A later `implement --from goal|plan` writes a fresh prd.json and flips `active:true` again. The `*-state.json` glob is deliberate. The hook blocks on any state f