brain-ecosystem

Get self-observation statistics: how Brain monitors its own behavior, performance metrics, and operational patterns.

Get self-observation insights: patterns Brain has detected about its own behavior and performance. Filter by type and limit results.

Get self-improvement plan: actionable suggestions Brain has generated based on observing its own performance patterns.

Get adaptive strategy status: all strategies with their current parameters, performance metrics, and revert rates.

View adaptation history: past parameter changes made by the adaptive strategy engine. Filter by strategy name.

Revert a specific strategy adaptation. Rolls back a parameter change and records the reason.

List all experiments. Filter by status to see running, completed, or aborted experiments.

Propose a new experiment with a hypothesis, independent/dependent variables, and control/treatment values. Brain will run it and collect statistical evidence.

Get detailed status of a specific experiment: progress, collected data points, preliminary results.

Get results of completed experiments with full statistical analysis: p-values, effect sizes, confidence intervals.

Abort a running experiment. Preserves any data collected so far.

Find correlations across different domains (errors, trades, posts, etc.). Reveals hidden connections between seemingly unrelated events.

Run a fresh cross-domain analysis. Scans recent events across all domains to discover new correlations and patterns.

Generate a human-readable narrative explaining cross-domain relationships. Tells the story of how different domains interact.

View past counterfactual analyses and their predictions vs actual outcomes.

Predict the impact of changing a variable to a new value. Estimates the effect on outcomes based on historical counterfactual data.

Get a summary of all distilled knowledge: principles, anti-patterns, and domain expertise that Brain has extracted from experience.

Get distilled principles: proven best practices and strategies that Brain has learned from data. Filter by domain.

Get distilled anti-patterns: known pitfalls, bad practices, and patterns to avoid. Learned from failures and negative outcomes.

Package distilled knowledge for a domain into a transferable format. Useful for sharing learnings between projects or brains.

View how knowledge has evolved over time. Shows how principles and anti-patterns have changed across periods.

View the current research agenda: prioritized list of open research questions and investigations Brain wants to pursue.

Get the next highest-priority research question to investigate. Use this to decide what Brain should focus on next.

Change the priority of a research question. Higher priority means it will be investigated sooner.

Add a new research question to the agenda. Brain will investigate it during research cycles.

List detected anomalies: unusual patterns, outliers, and unexpected behaviors that Brain has found. Filter by type.

Deep-dive into a specific anomaly: root cause analysis, related events, potential explanations, and recommended actions.

View anomaly detection history: past anomalies with their resolution status and outcomes.

Get a drift report: detect gradual changes in metrics, behavior patterns, or data distributions over time.

Read research journal entries: observations, discoveries, hypotheses, experiments, and reflections. Filter by type.

Get a summary of recent research journal activity: key themes, important discoveries, and active investigations.

View research milestones: significant achievements, breakthroughs, and key findings from the journal.

Write a new research journal entry. Document observations, discoveries, reflections, or experiment notes.

Search the research journal by keyword. Finds entries matching the query across titles, content, and tags.

Get Brain attention status: current work context, top topics by attention score, urgent topics, engine weights, and context switch history. Shows what Brain is paying attention to right now.

Manually direct Brain attention to a specific topic. Increases its attention score and urgency, making related research engines work harder on it.

Get the focus timeline — what topics Brain has been paying attention to over time, and what context switches happened.

Run the SelfMod benchmark suite — tests code generation quality across 10 tasks of varying difficulty. Returns pass/fail for each task with timing info.

Get the current benchmark score — shows pass rate of the latest run for each task, plus improvement trend over time.

List all available benchmark tasks with their difficulty and target files.

Get historical benchmark results to track improvement over time.

Start an autonomous browser task. The agent navigates, reads pages, fills forms, and extracts data using LLM-guided decisions.

Execute a scripted sequence of browser actions (navigate, click, type, screenshot, etc.).

Get the current status of the browser agent (running tasks, pages open, etc.).

Gracefully shut down the browser agent, closing all pages and releasing resources.

Ask the AI architecture critic a question about the Brain Ecosystem. Uses Claude API (Anthropic) by default, falls back to OpenAI. Maintains conversation history for multi-turn discussions.

Check AI critic bridge status: API keys configured, conversation history length.

Generate code using Claude API with Brain's accumulated knowledge (Principles, Anti-Patterns, Patterns from 22k repos) as context. The generated code stays in DB for review — no auto-deploy. Requires ANTHROPIC_API_KEY.

Overview of all code generations: total count, status distribution, tokens used, approval rate.

Approve or reject a code generation. Approved code gets journaled as discovery, rejected as failed experiment.

Show CodeMiner status: how many repos have been mined, content size, file breakdown.

Show extracted patterns from mined repos: top dependencies, tech stacks, project structures, README patterns. Use extract=true to re-analyze.

Get Brain concept abstraction status: total concepts, levels, transferability, top concepts.

View concept hierarchy tree for a specific concept, showing children and members.

Trigger concept formation NOW: clusters knowledge into abstract concepts with 3-level hierarchy.

List concepts with high cross-domain transferability — useful for knowledge transfer between brains.

Get Brain consciousness status: total thoughts, active engines, thought breakdown by engine/type/significance, and live dashboard client count.

Get recent thoughts from Brain consciousness stream. Optionally filter by engine name. Shows what Brain is thinking in real-time.

Store something in long-term conversation memory. Use for important facts, decisions, preferences, or context that should persist across sessions.

Semantically search long-term conversation memory. Returns memories ranked by relevance using RAG + FTS5.

Build a context summary from conversation memory for LLM consumption. Aggregates important recent memories.

Full-text search in conversation memory using FTS5.

Retrieve the most important memories, sorted by importance score.

Get conversation memory statistics: total memories, categories, storage info.

Get Brain curiosity status: knowledge gaps, unanswered questions, exploration rate (explore vs exploit), top bandit arms, and surprise count.

List Brain knowledge gaps: topics where attention is high but knowledge is low. Each gap has a type (dark_zone, shallow, contradictory, stale, unexplored), score, and auto-generated questions.

Ask Brain what to explore next. Uses UCB1 multi-armed bandit to balance exploration (new topics) vs exploitation (deepening known topics). Returns the recommended topic and suggested actions.

Detect surprises: things that violated Brain expectations. Finds confirmed-but-unlikely hypotheses, rejected-but-likely hypotheses, and experiments with unexpectedly large effects.

Start a multi-perspective debate on a question. Brain generates its perspective from principles, hypotheses, journal, anomalies, and predictions.

Synthesize a debate: compare all perspectives, detect conflicts, build weighted consensus, generate recommendations.

Generate Brain's perspective on a question without starting a full debate. Useful for cross-brain debates.

View past debates with their questions, perspectives, synthesis, and conflicts.

Challenge a principle by searching for contradicting evidence (Advocatus Diaboli). Returns resilience score and outcome (survived/weakened/disproved).

View recent principle challenges with their outcomes and resilience scores.

Show principles with the lowest resilience scores — the most vulnerable beliefs.

Synapse network health report: weight distribution, pruning history, regret analysis, diagnosis

Strategy emergence debug: why strategies are/aren't emerging from confirmed hypotheses

Fact extraction funnel: where dream-extracted facts get lost in the pipeline

Get Dream Mode status: whether dreaming is active, total cycles, lifetime consolidation stats. Dream Mode consolidates memories during idle periods — like sleep for Brain.

Manually trigger a Dream Mode consolidation cycle. Brain will replay memories, prune weak synapses, compress similar memories, and decay old unimportant ones.

Get past Dream Mode cycles. Shows what happened during each consolidation: replays, pruning, compression stats.

Get Brain emergence tracking status: emergent events, complexity metrics (Kolmogorov, Shannon entropy, Phi integration), surprise scores, metric trends.

Scan for emergent patterns NOW: unpredicted hypotheses, recurring anomalies without rules, cross-domain bridges, phase transitions, novel experiment effects.

Compute and return current complexity metrics: Kolmogorov complexity (compression ratio), Shannon entropy, network density, knowledge diversity, integration Phi.

View the emergence journal: all emergent events sorted by surprise score, with evidence and source engines.

Get the current emotional state of the brain — mood, dimensions (frustration, curiosity, surprise, confidence, satisfaction, stress, momentum, creativity), and overall status.

Get the emotional history of the brain over time — shows how mood and dimensions changed across cycles.

See what caused recent mood changes — which engines triggered emotional dimension shifts.

Get behavior recommendations based on the current emotional state — what should the brain do given its mood?

Get Brain evolution engine status: current generation, population size, best fitness, champion genome, config.

Get Brain evolution generation history — fitness trends across generations.

Get the best-performing individual (champion) from Brain evolution — genome, fitness, lineage.

Manually trigger a Brain evolution generation — tournament selection, crossover, mutation, fitness evaluation.

Show experiment ledger status — active experiments, kept/reverted counts

Start a controlled A/B experiment for a system change

Show experiment history — past experiments with decisions

Get Brain goal tracking status: active/achieved/failed goals, recent achievements, top active goals.

Create a new goal for Brain to track and pursue autonomously. Brain will monitor the metric and report progress.

Get detailed progress report for a specific Brain goal, including trend analysis and completion forecast.

Search Brain knowledge base using vector similarity (RAG). Searches across insights, memories, principles, errors, solutions, rules.