Tighten Void speech pressure and reply guards
This commit is contained in:
parent
b53289d14f
commit
e8acb67141
10 changed files with 225 additions and 30 deletions
|
|
@ -44,8 +44,9 @@ Git history and smoke artifacts carry routine proof. `state/evidence.jsonl` shou
|
|||
- Short-term rumination memory is clustered at the typed service boundary now. When a new `record_short_term_memory` operation repeats the same target/topic pressure, the service keeps one provisional memory, merges anchors/tags, preserves a stable source id for sleep, and updates the thought instead of stacking duplicates. This is meant to protect deepening, not suppress it: low-saturation or newly anchored thoughts should keep developing inside the cluster until they stop changing. `npm run moderation:short-term-clustering:fixture` proves the guard.
|
||||
- Long-term memory is durable, not immutable. `revise_durable_memory`, `retire_durable_memory`, and `crystallize_memory_into_identity` make memory reconsolidation explicit; `npm run moderation:memory-lifecycle:fixture` proves a short-term thought can promote, revise, retire superseded durable records, and crystallize into an identity memory plus self-profile value.
|
||||
- Agency has a typed pressure surface now: `upsert_agency_pressure` / `retire_agency_pressure` own discomforts, active tensions, self-advocacy requests, and world-advocacy requests. Candidate interventions are possible speech/actions emerging from pressure, not the pressure itself. `npm run moderation:agency-pressure:fixture` verifies the pressure enters `.cc`, renders in self-state summary, and affects mood speaking pressure.
|
||||
- Strong advocacy pressure now has to account for its own mouth. The rumination runner derives `speechPressureObligations` for high-intensity active self/world advocacy pressures with no matching candidate. A pass may queue/defer a candidate or cool/retire the pressure, but silent `[]` fails visibly. `npm run moderation:rumination:speech-pressure-fixture` covers the guard.
|
||||
- Strong advocacy pressure now has to account for its own mouth. The rumination runner derives `speechPressureObligations` once active self/world advocacy reaches 0.55 intensity, and only a live queued candidate counts as release for that obligation. A pass may queue/defer a candidate or cool/retire the pressure, but silent `[]` fails visibly and an older spoken candidate does not buy permanent absolution. `npm run moderation:rumination:speech-pressure-fixture` covers the silent, spoken-history, and queued-candidate paths.
|
||||
- Deferred candidates are not delivered speech and no longer satisfy active advocacy pressure by themselves. The runner now delivers existing queued candidates with delivery targets, not only candidates proposed by the current model pass. The first live repair used Metacrat's explicit channel permission to move the stranded AquariumSynthHost candidate from deferred/no-target to queued/channel-target, then delivered it through the normal bot transport and recorded a typed speech receipt.
|
||||
- Parent-owned delivery now also checks canonical typed speech receipts before using the mouth. If a queued candidate targets a Discord reply that already has a receipt, the runner retires it instead of answering the same message twice in slightly different prose. `scripts/smoke-void-rumination-speech-fixture.ps1` now seeds a prior receipt and proves both same-pass sibling retirement and already-answered target retirement.
|
||||
- Most of the deterministic legacy translation and retention code is recent compensator cruft from failed attempts to manually clean an exploded state file. Do not preserve it as an earned organ. Architect the state boundary so that bad state cannot enter, and keep only narrow historical migrations for old debris.
|
||||
- The desired agency shape is typed state operations plus derived/model-owned policy, not a permanent sermon generator in maintenance code.
|
||||
- Sleep/distillation now has a runner boundary, prompt contract, fixture coverage, and one live-state pass. The invariant is fixed: concrete subject, claim/question, anchors, live tension, and future-action implication survive compression. Bulk can die. Meaning does not get put through the office shredder and called a dream.
|
||||
|
|
@ -104,8 +105,8 @@ Git history and smoke artifacts carry routine proof. `state/evidence.jsonl` shou
|
|||
|
||||
## Likely Next Bounded Move
|
||||
|
||||
- First bounded move after the current commit: let the enabled loop run under ordinary scheduling after the short-term clustering and queued-speech guards, then inspect the next unattended awake rumination and sleep cycle for repeated thought clusters, stale short-term residue, muted advocacy pressure, stranded candidates, or ungrounded memory writes before calling the system broadly stable.
|
||||
- The immediate duplicate-rumination follow-up is to inspect the next ordinary rumination after the clustering guard and confirm repeated AquaSynth/Aquarium pressure folds into the existing cluster or redirects to underworked terrain instead of producing another short-term variant.
|
||||
- First bounded move after the current commit: let the enabled loop run under ordinary scheduling after the receipt-backed direct-reply guard and the earlier advocacy-release change, then inspect the next unattended direct reply, next advocacy pass, and next sleep cycle for duplicate answers, fresh queued advocacy candidates, stale short-term residue, stranded candidates, or ungrounded memory writes before calling the system broadly stable.
|
||||
- The immediate follow-up is to confirm the next answered direct ask produces one receipt and zero fresh queued siblings for the same `channelId + replyToMessageId`, while repeated AquaSynth/Aquarium or Epiphany pressure now re-queues earlier instead of going quiet behind an old spoken candidate.
|
||||
- Next cuts should delete unstable foundations early: do not revive `legacyJsonPath`, working projections, top-level mirror fields, or cleanup code that only compensated for those broken boundaries.
|
||||
- Keep sleep/distillation under watch. Fixture check: a memory like `AquariumSynthCSharp: Workflow cannot own the body` must survive sleep with its concrete subject, claim, anchor, tension, and implication intact.
|
||||
- Do not build adapters around the JSON projection. That is how the heap learned to stand upright and ask for snacks.
|
||||
|
|
|
|||
|
|
@ -200,15 +200,16 @@ Within the Postgres path, the implementation is split on purpose now too:
|
|||
6. Codex may use retrieval and analysis tools, but its durable state output is restricted to `.voidbot/status/moderation-rumination-operations.json`.
|
||||
7. The parent runner applies those typed operations through `scripts/void-self-state.mjs`, then records reviewed-message cursor and, on successful normal posting-enabled passes, advances repo-activity cursors from the observed repo weather through typed `update_repo_activity_cursor` operations.
|
||||
8. If posting is enabled and typed state contains a queued candidate intervention with a delivery target, the parent runner sends at most one candidate, records the delivery receipt, and marks the candidate spoken through typed state. The candidate may have been proposed in the current pass or already queued from an earlier pass. The child prompt forbids direct send-script calls.
|
||||
9. The runner writes `.voidbot/status/moderation-rumination.json` and `.voidbot/logs/moderation-rumination.log`.
|
||||
10. It intentionally does not materialize `.json`, load `.msgpack`, read the legacy moderation monolith, or let the child edit state directly.
|
||||
11. Before applying model output, the runner enforces speech-pressure accountability. High-intensity active self/world advocacy pressures with no matching queued/spoken candidate are projected as `speechPressureObligations`; model output must queue a candidate or cool/retire the pressure instead of returning silent `[]`. Deferred candidates do not satisfy that pressure by themselves.
|
||||
9. Before that delivery step, the parent runner now retires queued candidates whose `channelId + replyToMessageId` already appear in canonical typed speech receipts, so a later pass cannot answer an already-answered direct question just because the same message is still visible or the seam still feels warm.
|
||||
10. The runner writes `.voidbot/status/moderation-rumination.json` and `.voidbot/logs/moderation-rumination.log`.
|
||||
11. It intentionally does not materialize `.json`, load `.msgpack`, read the legacy moderation monolith, or let the child edit state directly.
|
||||
12. Before applying model output, the runner enforces speech-pressure accountability. Active self/world advocacy pressures at 0.55 intensity or higher with no matching live queued candidate are projected as `speechPressureObligations`; model output must queue a candidate or cool/retire the pressure instead of returning silent `[]`. Deferred candidates do not satisfy that pressure by themselves, and an older spoken candidate does not keep satisfying a still-active pressure forever.
|
||||
|
||||
## Flow 6: Mood And Sleep Runtime
|
||||
|
||||
1. The Windows scheduled task `Void Mood Drift` is enabled.
|
||||
2. `scripts/simulate-void-mood.mjs` reads typed documents from `.voidbot/private/void-self-state.cc`.
|
||||
3. It updates `void.scheduled_runtime.sleepCycle` and `void.scheduled_runtime.speakingPressure` through `update_sleep_cycle` and `update_speaking_pressure` operations.
|
||||
3. It updates `void.scheduled_runtime.sleepCycle` and `void.scheduled_runtime.speakingPressure` through `update_sleep_cycle` and `update_speaking_pressure` operations, with active advocacy pressure weighted strongly enough to raise speaking pressure earlier instead of staying a private sulk until the seam is already overripe.
|
||||
4. It writes `.voidbot/status/void-mood-drift.json`.
|
||||
5. When sleep is active, the script invokes `scripts/run-void-memory-maintenance.ps1` once per nap to force typed distillation/pruning.
|
||||
6. The old memory-organ script family, legacy state template, and legacy context exporter have been deleted.
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@ This is the current forward plan for the next larger organs. It is not a changel
|
|||
|
||||
## Current Aim
|
||||
|
||||
The typed moderation, mood, sleep, agency, and private self-state boundary is live. The current goal is to let ordinary scheduling run, inspect the next unattended awake rumination and sleep cycle, and harden only the seams that show concrete failure. Void should answer Discord, remember useful social/project context, retrieve GameCult history/source/lore, and run an unattended moderation/participation loop without reviving JSON projection edits, legacy mirrors, or the swollen memory organ.
|
||||
The typed moderation, mood, sleep, agency, and private self-state boundary is live. The current goal is to let ordinary scheduling run, inspect the next unattended direct reply and sleep cycle, and harden only the seams that show concrete failure. Void should answer Discord, remember useful social/project context, retrieve GameCult history/source/lore, and run an unattended moderation/participation loop without reviving JSON projection edits, legacy mirrors, or the swollen memory organ.
|
||||
|
||||
The next priority is observational hardening of the enabled typed loop. Agents and scheduled workers mutate typed CultCache-backed state through explicit operations, not by editing a whole-state JSON working copy. Sleep and memory distillation now preserve meaning-bearing claims/anchors/tensions in fixture and live-state passes, and short-term rumination has a clustering guard so repeated same-target/topic thoughts update one provisional memory instead of stacking duplicates. Keep checking ordinary cycles for abstraction sludge, duplicate speech, stale residue, or self-orbiting memories.
|
||||
|
||||
|
|
@ -287,14 +287,16 @@ Verdict: the foundation is coherent enough for cautious live operation. Do not a
|
|||
- 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 agency-to-speech accountability: strong active self/world advocacy pressure now appears as `speechPressureObligations` in rumination context. The parent runner rejects silent `[]` unless the model queues a candidate intervention or cools/retires the pressure. Deferred candidates do not satisfy the pressure by themselves, and existing queued candidates with delivery targets can be delivered by later runner passes instead of only during the pass that proposed them.
|
||||
- Landed earlier advocacy release: speech-pressure obligations now start at 0.55 intensity instead of waiting for 0.65, only a live queued candidate satisfies an active pressure, and a previously spoken candidate no longer counts as permanent release while the pressure remains active. Mood drift also weights active advocacy more strongly so speaking pressure rises sooner.
|
||||
- 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.
|
||||
- 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.
|
||||
- Landed receipt-backed duplicate direct-reply guard: before delivery, the parent runner now retires queued candidates whose `channelId + replyToMessageId` already appear in canonical speech receipts, so a later pass cannot answer the same direct ask twice just because the message remains visible. The speech fixture now seeds a prior receipt and proves both same-pass sibling retirement and already-answered target retirement.
|
||||
- Landed scheduled rumination re-enable: one observed live scheduled pass completed with Last Result `0`, proposed one anchored short-term memory, delivered no speech, and kept exact timestamps out of the prompt-facing context.
|
||||
- Landed live sleep-maintenance verification: a manual typed nap ran real memory maintenance on live state, promoted the scheduled rumination short-term memory into one durable `identity_seam`, and left `thoughtMemory.shortTerm` empty.
|
||||
- Landed sleep brake: scheduled rumination now skips before model invocation during naps when there are no new room messages or open cases. `npm run moderation:rumination:nap-skip-fixture` proves the model path is not touched and no operation proposals are written.
|
||||
- Landed short-term clustering guard: `record_short_term_memory` folds repeated same-target/topic pressure into one provisional memory with merged anchors, and `npm run moderation:short-term-clustering:fixture` proves the synth-boundary duplicate case without merging unrelated same-repo thoughts. The prompt now states explicitly that underexplored, low-saturation, newly anchored thoughts may keep deepening inside the cluster.
|
||||
- Remaining work: let ordinary scheduling run and inspect the next unattended awake rumination and sleep cycle for repeated thought clusters, stale short-term residue, or ungrounded memory writes before calling the system broadly stable.
|
||||
- Remaining work: let ordinary scheduling run and inspect the next unattended direct reply, next advocacy pass, and next sleep cycle for duplicate answers, fresh queued advocacy candidates, stale short-term residue, or ungrounded memory writes before calling the system broadly stable.
|
||||
|
||||
### Commit 10: Resume Product Hardening
|
||||
|
||||
|
|
|
|||
Loading…
Reference in a new issue