Documentation Index
Fetch the complete documentation index at: https://docs.bumbleagi.com/llms.txt
Use this file to discover all available pages before exploring further.
GEN (Generative Entropic Noise) is the small-model stream that fills the ## Noise section of the body the entity reads. In code: NoiseEngine inside TonicBody (bumblebee/identity/soma.py).
This page is the data-flow companion to Soma. Read that page for tuning and philosophy; read this one to see exactly what hits the noise LLM.
GEN is not the main model’s chain-of-thought. It does not receive raw tool outputs, full system prompts, or the entire chat log. It does not read body.md as input — it writes fragments that appear when the body is rendered and flushed.
When noise runs
TonicBody.maybe_tick_noise calls NoiseEngine.generate only if noise is enabled and noise.should_tick() is true (wall time since the last tick ≥ soma.noise.cycle_seconds).
Call sites:
- Presence daemon heartbeat (
bumblebee/presence/daemon.py) — builds journal_tail + conversation_tail from the live entity.
- After each committed turn (
Entity._tick_noise_post_turn in bumblebee/entity.py) — the noise clock is reset so GEN can refresh during active chat.
Ebb: If soma.ebb is on with skip_post_turn_noise_when_quiet: true, post-turn regeneration may be skipped when the presentation tier is quiet (low salience).
What is passed into each generate call
| Input | Source |
|---|
| Bars summary | One line of current bar percentages (snapshot_pct()). |
| Affects summary | Rendered current affects (from the separate affect LLM pass, not from noise). |
| Recent events | Last 5 entries on TonicBody._recent_events. |
| Journal tail | Last ~800 characters of journal.md if present. |
| Conversation tail | Last 8 non-system messages from Entity._history, ~500 chars each (user: / assistant:). |
| Entity name | Config name. |
| Prior noise | Last 4 fragments — “do not repeat” context. |
| Shape hint | One random line from _NOISE_SHAPE_HINTS (large catalog: mundane pings, sensory beats, schedule thoughts, explicit anti-metaphor nudges, etc.). |
soma.noise.model if set, else reflex/deliberate fallback. Temperature and max_tokens come from soma.noise (default max_tokens is sized for 2–7 short fragments per tick).
Exogenous seed sources (NoiseSeeder)
Before GEN generates fragments, the NoiseSeeder (bumblebee/identity/noise_seeder.py) selects one exogenous seed per tick from a weighted source pool. These seeds bias the noise toward specific associative domains:
| Source key | What it produces |
|---|
episodic_random | A random old episode summary — memory resurfacing |
belief_random | A stored belief with confidence score |
knowledge_random | A section from the entity’s knowledge file |
world_discovery | A concept from the concept corpus or Wikipedia, with associative chaining |
relationship_echo | A relational document tail — thinking about someone |
journal_echo | A snippet from the entity’s journal |
temporal | Time-of-day or day-of-week contextual seed |
counterfactual_simulation | New. Picks a past episode and frames it as “what if I had acted differently?” — self-reflection on past actions |
dream_state | New. Surreal cross-domain association (e.g. “how does cooking relate to physics?”) — activates during dormant periods |
web_venturing | New. Explicit seed urging the agent to follow a curiosity onto the internet — search, read articles, learn something new |
Mood-congruent daydreaming
Seed source weights are dynamically biased by the current Soma state. The NoiseSeeder._suppress_weights() method reads bar percentages from tonic.bars.snapshot_pct() and adjusts weights:
- High tension (>75%) → 2×
episodic_random (rumination), 0.2× world_discovery (agent turns inward)
- Low social / high loneliness → 2.5×
relationship_echo (agent thinks about people)
- High curiosity (>70%) → 2×
web_venturing, 1.5× world_discovery (agent looks outward)
This creates a tight feedback loop: the body’s felt state directly shapes what the agent daydreams about.
Associative chaining
The world_discovery source now implements associative chaining instead of pure random concept selection. After picking a concept, it stores the concept in _last_concept_thread. On the next tick (70% of the time), it scores candidate concepts by lexical overlap with the previous concept and preferentially picks from the top-5 most related. This creates daydream-like threads where one thought naturally leads to another, rather than disconnected jumps.
Web venturing and autonomous exploration
When web_venturing fires as a seed source, and the wake cycle detects this in the GEN fragment buffer, the wake engine auto-escalates to wide mode — giving the agent more rounds and tool budget to actually follow through on the curiosity. The salience bias block also injects explicit encouragement to use search_web, fetch_url, and other tools to explore the real internet.
This is the primary mechanism by which the agent’s internal state drives it to autonomously venture out into the world.
Intuition: noise riffs on how the body feels + a thin event strip + diary scrap + chat tail — so recent themes (lots of web/tools) show up via event names and history, while raw API prose usually only appears if it is already in chat or journal.
Recent events (what _recent_events contains)
Formatted for the prompt by _format_event_for_noise:
| Event | Typical meaning |
|---|
message_received | Turn start — who, length, platform/channel. |
message_sent | After commit — recipient, length, platform. |
action | Each tool call — tool name + ok / error (not full output). |
idle | Daemon — long silence, minutes. |
mood_declared | End-turn mood from tool state, if any. |
world_poke | External cue via poke_world (TTL’d). |
Generation behavior (prompting)
- Batch size: each completion yields 2–7 parsed fragments (newlines / blank lines), then capped before merging into the rolling deque (
max_fragments still limits total buffer size).
- Voice: prompts push uneven subconscious scraps and discourage one long metaphorical monologue (including repeated “tech spirituality” clichés) unless a shape hint steers otherwise.
- Short lines: fragments can be very short (minimum length after sanitization is low) so spikes like
ok or hm can survive if the model emits them.
- Shape pressure: one random instruction per tick steers form (e.g. very short blunt line, no questions, sensory-only, “avoid API/map/ink imagery this batch”). See
_NOISE_SHAPE_HINTS in bumblebee/identity/soma.py.
Ebb is orthogonal: it only changes how much rendered noise appears in the main perceive prompt (quiet / normal / high) and whether post-turn regeneration is skipped — not the generation rules above.
Code pointers
| Piece | Location |
|---|
| Noise LLM + shape hints | NoiseEngine.generate, _NOISE_SHAPE_HINTS — bumblebee/identity/soma.py |
| Gate + bars/affects/events assembly | TonicBody.maybe_tick_noise — same file |
| Event log | emit, apply_immediate, _recent_events — same file |
| Exogenous seeds | NoiseSeeder — bumblebee/identity/noise_seeder.py |
| Daemon | PresenceDaemon heartbeat — bumblebee/presence/daemon.py (_build_conversation_tail) |
| Post-turn | Entity._tick_noise_post_turn — bumblebee/entity.py |
| Ebb skip | should_skip_post_turn_noise — bumblebee/identity/soma.py |
| Wake auto-escalation | _gen_has_web_venturing, _soma_curiosity_high — bumblebee/presence/wake_cycle.py |
- Soma — bars, affects, GEN overview, ebb,
body.md, configuration.
- Telegram guide — busy indicator (harness UX, separate from GEN).
