GameCult/VoidBot
Watch
9
0
Fork
You've already forked VoidBot
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Move rumination speech under parent control

Browse source
This commit is contained in:
Meta Vens 2026-05-17 13:56:57 +01:00
commit d154aabc69
13 changed files with 374 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

6
notes/voidbot-implementation-plan.md
View file

@ -239,7 +239,7 @@ Verdict: stop feature work and rebuild this foundation. The rest of the machine
### Commit 6.5: Keep Old Scripts Offline
- `Void Mood Drift` has been deliberately reinstalled after the typed sleep-maintenance fixture passed. It runs `scripts/run-void-mood-drift.ps1`, which calls the typed `scripts/simulate-void-mood.mjs` path.
- Keep `Void Moderator Rumination` disabled until its typed operation path has passed a controlled real-model no-post run and review. The model-branch fixture now proves runner plumbing; it does not prove the live model's judgment is ready for unattended speech.
- Re-enable `Void Moderator Rumination` only after typed operation quality and parent-owned speech closure have passed. The runner now has both a model-branch fixture and a parent-owned speech fixture; the first live scheduled pass still needs inspection before the loop is treated as stable.
- Do not reinstall or re-enable deleted legacy scripts as compatibility surfaces.
- Verification: scheduled-task query shows mood enabled with Last Result `0`, moderation disabled, and rebuilt direct dry runs operate only on `.voidbot/private/void-self-state.cc`.
@ -283,7 +283,9 @@ Verdict: stop feature work and rebuild this foundation. The rest of the machine
- Landed agency ingress: `upsert_agency_pressure` / `retire_agency_pressure` store discomforts, active tensions, self-advocacy requests, and world-advocacy requests separately from candidate speech.
- Landed agency verification: `npm run moderation:agency-pressure:fixture` proves agency pressure enters typed state, appears in the self-state summary, and contributes to mood speaking pressure.
- Landed rumination verification: `npm run moderation:rumination:fixture` routes the non-skip rumination branch through a fake Codex child and proves short-term memory, incubation, agency-pressure, and candidate-intervention proposals cross the parent-applied typed operation boundary without timestamp leakage.
- Remaining work: run a controlled real-model typed rumination pass with `-NoPost`, inspect whether the proposed operations are concrete enough to survive sleep, and decide whether unattended scheduling is ready or whether action/receipt closure still needs hardening.
- Landed controlled live rumination review: the first real-model `-NoPost` pass produced valid but self-orbiting readiness-test residue, which was pruned/retired and folded into a prompt guard; the second real-model `-NoPost` pass sampled non-VoidBot repo pressure, used retrieval, and correctly wrote `[]` when nothing earned persistence.
- Landed parent-owned speech closure: candidate interventions can include `deliveryTarget`, the prompt forbids child-side send scripts, the parent runner delivers at most one newly queued deliverable candidate per pass, records the receipt, and marks the candidate spoken with `mark_candidate_intervention_spoken`. `npm run moderation:rumination:speech-fixture` proves the loop with fake Codex and fake Discord send helpers.
- Remaining work: re-enable the scheduled task cautiously, observe one scheduled pass, and inspect operation quality plus typed state before calling the loop stable.
### Commit 10: Resume Product Hardening