This note is the source-grounded description of how the live VoidBot stack is shaped right now. It is not a prose replacement for the code, and it is not a second canonical map. Its job is to explain the major organs and their data flow in plain language with concrete path anchors.
- request context assembly, including interaction-memory/profile attachment, shared Void self-state projection, and any precomputed situational social read.
- quick Ollama sidecar inferer for ephemeral room-reading scaffolding built from the current prompt, recent room transcript, and longer-horizon interaction memory.
- archived Discord message store under `.voidbot/rag/messages.json`, including bot-directed prompts that stay preserved for forensics even when they are excluded from default history retrieval.
1.`apps/bot/src/discord-bot.ts` receives a slash command or mention-driven request and delegates prompt/command work into `apps/bot/src/discord-bot-handlers.ts`.
5.`packages/core/src/interaction-memory-analysis.ts` turns direct prompts or ambient mentions into normalized remembered events; `packages/core/src/interaction-memory-profile.ts` distills remembered events into a reusable longer-horizon social read.
6.`apps/bot/src/discord-bot-handlers.ts` usually runs a quick `think=false` local Ollama sidecar pass over the prompt and recent room transcript to infer an ephemeral situational social read, then records that read as an audit event for later aggregation. If the shared self-state says Void is napping, the handler skips that extra room-read pass and keeps the request deliberately cheap.
7.`packages/core/src/context-builder.ts` assembles request context, including recent interaction profile, the situational social read, the shared Void self-state projection, and any retrieval hints.
8. Provider selection goes through `packages/providers/src/index.ts`, but the handler can downshift owner traffic into the cheaper local lane when the shared self-state says Void is napping.
3.`packages/providers/src/owner-codex-provider.ts` runs `codex exec` in the bounded lane, exposes the VoidBot MCP server, records traces, and carries advisory source-grounding hints plus retrieval tools for repo/lore/project questions.
Runtime and parsing live in `packages/providers/src/owner-codex-runtime.ts`; prompt and artifact rendering live in `packages/providers/src/owner-codex-render.ts`.
2. Bot-directed prompts are tagged at ingest, kept in the raw Discord archive, and deliberately skipped when the history ingester builds semantic chunks so repeated summons stop poisoning normal retrieval.
3.`packages/rag/src/retrieval-service.ts` translates history/source queries into vector lookups plus metadata filters.
4.`packages/rag/src/qdrant-vector-store.ts` executes the live vector lookups against separate history and source collections.
5.`scripts/reconcile-source-repos.ts` discovers local Git repos, refreshes push hooks, prunes stale repo shards, and indexes newly discovered repos.
6.`scripts/index-source-repos.ts` and `scripts/git-post-push-index.mjs` drive explicit or detached per-repo source/lore reindex work.
7.`apps/worker/src/mcp-server.ts` boots the MCP lane, while `mcp-server-resources.ts` and `mcp-server-tools.ts` expose retrieval to Codex and other sessions through `search_history`, `get_message_context`, `search_sources`, `get_source_context`, and `list_indexed_repos`.
- scheduled moderation/participation runner that launches `codex exec` from the VoidBot workspace with the usual MCP/tool surface, writes status/log pulse files, and lets the unattended loop edit only the JSON working projection while the canonical moderation state lives in the sibling MessagePack cache file.
-`scripts/moderation-state-store.mjs`
- tiny Node wrapper around the built moderation-state cache helper; used by PowerShell and Node scripts to materialize the JSON working projection, commit it back into canonical MessagePack, and update cursors without re-implementing persistence semantics in every script.
- plain Node mood-drift organ that nudges the shared personality-vector activations with Perlin-shaped noise, damps them back toward their means using plasticity, updates the speak/confession/novelty pressure meter in the shared self-state, maintains the nap-cycle clock plus sleep-state vector bias, and now also reconciles compact semantic-memory vectors plus incubation/dream state.
-`scripts/void-memory-organ.mjs`
- hybrid symbolic/vector memory helper used by mood drift to attach compact semantic vectors to persistent memories and runtime seeds, build resonance clusters, keep an incubation queue alive across turns, and leave more meaningful dream residue during naps.
- hidden-task wrapper for the mood-drift organ; writes status/log pulse files and respects the moderation lock so the two state writers do not stomp each other for sport.
-`scripts/install-void-mood-drift-task.ps1`
- installs the local 5-minute scheduled task that keeps the shared self-state emotionally non-flat between moderation passes.
- plain Node helper for novelty excursions that samples one random archived Discord seam, returns a local context window around the anchor message, and skips bot-directed prompt rows by default.
When present, an ignored local script at `.voidbot/private/check-voidbot-operations.local.ps1` can inject adjacent-service checks into the same ops lane. That keeps private infrastructure gossip out of tracked code while preserving one boring place for smoke alarms to land.
The important scar is that these stores are split on purpose. Do not casually weld them back into one convenient blob and act surprised when it becomes a moon.
1. The Windows scheduled task `Void Moderator Rumination` starts `scripts/run-void-moderator-rumination.ps1` every 15 minutes.
2. The runner launches `codex exec` from `E:\Projects\VoidBot` with full local workspace/tool access so the child agent sees the same project rules, `.codex/config.toml`, and `voidbot` MCP server as a normal VoidBot workspace session.
3. The child agent reads `config/discord-server-rules.md`, `config/moderation-review-agent.md`, `styles/void-default.md`, and the editable working projection `.voidbot/private/moderation-agent-state.json`. The runner materializes that projection from the canonical `.voidbot/private/moderation-agent-state.msgpack` file before each pass and commits it back afterward.
4. For chronological heartbeat duty, it polls `node scripts/export-recent-discord-history.mjs`; when the room is quiet and incubation feels hungry, stale, or oversaturated, it can use `node scripts/export-random-discord-history.mjs` to widen the branch. If an incubating thought is already live and grounded, the loop is allowed to deepen that thought directly instead of performing a novelty errand for ceremony's sake. It also uses `node scripts/export-recent-repo-activity.mjs --hours 96 --max-commits 3 --cursor-file .voidbot/private/moderation-agent-state.json` when repo motion is the actual seam or when the queue needs fresher fuel; that helper advances `moderation_runtime.repo_activity_cursor` so the same four-day commit window does not keep getting reinjected as fresh weather. The loop can inspect exact diffs or nearby source context when one repo seam looks alive, and uses semantic `search_history` to novelty-check candidate thoughts before deciding whether they are already in the room.
Quiet awake passes are now supposed to philosophize from the surviving seam first: ask what governing principle the thought is circling, what tension keeps it from collapsing into slogan, and what embodiment consequence would make it real. Retrieval is meant to support that work when the seam is hungry or contradictory, not replace it with endless specimen collection.
The doctrine now also biases that philosophical pass back toward concrete machines. If a seam stays too abstract, the loop is supposed to cash it out in a specific repo, subsystem, gameplay mechanic, lore structure, frontend surface, or agent organ before letting it dominate another run.
5. If speaking would genuinely help, it posts through `node scripts/send-discord-message.mjs`; for guild channels and threads that helper can either speak as the base bot or execute through the shared per-channel webhook pipe with an overridden persona name/avatar so swarm agents can front without needing separate Discord bot identities. Otherwise it updates only the ignored moderation-state working projection, which the runner then commits back into canonical MessagePack with the typed cache helper. The state carries cursors, open room obligations, memories, analytic/associative thought lanes, bridge syntheses and saturation notes, recent archive excursions, recent repo-activity sweeps, novelty checks, speaking-bias hints, and candidate interventions. A direct ask aimed at Void is supposed to survive as an `open_cases` obligation until it is answered or explicitly retired; cursor advancement alone is not closure.
6. The runner writes pulse files under `.voidbot/status/moderation-rumination.json` plus `.voidbot/logs/moderation-rumination*`, so the loop can be observed without pretending an app heartbeat card is a real workstation daemon.
2. That wrapper calls `node scripts/simulate-void-mood.mjs`, which reads canonical moderation state from `.voidbot/private/moderation-agent-state.msgpack`, then refreshes the editable JSON working projection after each pass.
3. The script first runs `scripts/void-memory-organ.mjs` over the shared self-state: episodic/semantic/musing/dream memories plus recent repo/archive/runtime seeds get compact semantic vectors, the strongest cross-memory similarities become `moderation_runtime.memory_resonance`, and the best cluster candidates are promoted into `moderation_runtime.incubation`.
The collector now intentionally downweights repetitive empty-room bookkeeping, so strings like `no new messages / no smoke / no post` stop masquerading as the deepest thought in the machine just because the room stayed quiet for a while.
Incubation scoring now distinguishes novelty-to-room from novelty-to-self, merges repeated seams into existing support instead of minting fresh shards, and writes `bridge.source_coverage` plus `bridge.refractory_topics` so a thought that keeps winning can be cooled off in favor of less-worked repos, archive years, or channels.
Cluster labeling is no longer supposed to be a slash-joined keyword graveyard. The organ now tries to synthesize each cluster into a claim, question, or fascination target with an explicit target kind such as repo, archive, room, self, or system, and uses concreteness/fertility/cross-domain pressure to decide whether the thought deserves more curiosity.
When the sleep cycle says Void is napping, that same organ becomes deliberately lossy: it promotes a few stronger incubating seams into semantic memory, then trims raw episodic receipts, repo sweeps, archive excursions, novelty checks, and recent musings back down so the next awake pass has to think from carved seams rather than yesterday's full notebook.
The same organ can also crystallize an over-chewed private seam into identity: once a thought has survived enough passes to become durable self-doctrine rather than live curiosity, it can be retired from the active queue, copied into `identity.private_notes`, echoed as a canonical value, preserved as one `identity_seam` semantic memory, and queued as a warmer `identity_crystallization` candidate intervention so the room has a chance to hear it. Those intervention records can be marked `mustEventuallyShare`, which ordinary pruning is supposed to preserve until the room gets a decent opening.
4. It then nudges each canonical-state `current_activation` toward a Perlin-shaped target centered on the vector mean, with excursion size and drift rate governed by plasticity and category-specific tempo.
5. The same script also maintains `moderation_runtime.sleep_cycle`: every four hours it opens a one-hour nap window, marks the shared self-state as napping or awake, and adds a small sleep-bias push toward exhaustion, figurative language, self-disclosure, and reduced technical density while the nap is active. During those nap windows it can also create or reinforce distilled dream residue from the strongest incubating thought, but the main job of sleep is memory cleanup and seam consolidation rather than awake-style philosophizing.
6. It also reads `.voidbot/status/void-last-speech.json` so recent actual speech damps `needToSpeak` instead of letting the herald impulse climb forever like a busted pressure gauge.
7. The updated `speaking_bias` block, sleep-cycle projection, incubation/resonance state, and shifted activations are written back into the shared self-state, and a pulse lands in `.voidbot/status/void-mood-drift.json`.
