This document is the map of the reinforcement-learning surface in<\/p>\n\n\n\n
_AugmentedIntelligence. It explains the shared RL infrastructure, the domain<\/p>\n\n\n\n
modules, the training\/deployment ladder, and the commands used to inspect or run<\/p>\n\n\n\n
the systems.<\/p>\n\n\n\n
For MathNN details, see docs\/MathNN.md. For the live curriculum catalog, see<\/p>\n\n\n\n
docs\/RLCurriculum.md.<\/p>\n\n\n\n
The RL stack has three layers:<\/p>\n\n\n\n
algorithms, exploration, planning, neural DQN, distributed RL, and deployment<\/p>\n\n\n\n
gates.<\/p>\n\n\n\n
parenting, political-news triage, real-world hooks, and curriculum smoke<\/p>\n\n\n\n
environments.<\/p>\n\n\n\n
shadow mode, gated live execution, safe-action shields, checkpoints, and<\/p>\n\n\n\n
promotion rules.<\/p>\n\n\n\n
The preferred path for any domain is:<\/p>\n\n\n\n
[text] Do not skip directly to live actuation. High-stakes domains should remain in<\/p>\n\n\n\n simulation, offline analysis, or human-reviewed advisory mode unless there is a<\/p>\n\n\n\n separate validated safety layer.<\/p>\n\n\n\n These run locally without a tensor\/autograd backend:<\/p>\n\n\n\n Use these for smoke tests, tabular policies, smaller state vectors, action<\/p>\n\n\n\n selection, and local DQN experiments.<\/p>\n\n\n\n These are declared in the deep-RL interface and route to a remote policy server<\/p>\n\n\n\n or tensor\/autograd backend for training:<\/p>\n\n\n\n Use these when the policy needs large neural networks, continuous-control<\/p>\n\n\n\n training, offline deep RL, imitation learning, or heavy world-model training.<\/p>\n\n\n\n General RL:<\/p>\n\n\n\n [text] Neural per-agent policy registry:<\/p>\n\n\n\n [text] Real-world\/deployment:<\/p>\n\n\n\n [text] Curriculum:<\/p>\n\n\n\n [text] Short aliases:<\/p>\n\n\n\n [text] Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Minecraft DQN observes player\/server state through the Spigot bridge, selects<\/p>\n\n\n\n actions, computes rewards, logs transitions, and can execute actions through the<\/p>\n\n\n\n server bridge, local client input, or both.<\/p>\n\n\n\n State\/action shape:<\/p>\n\n\n\n inventory\/tool flags, blocked directions, target block information, world,<\/p>\n\n\n\n biome, game mode, selected item, and bridge observation ID.<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Client-control commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Start in observe and dry-run, enable wall-guard and unstuck, then train<\/p>\n\n\n\n in short sessions. Use minecraft dqn net on when you want the shared MathNN<\/p>\n\n\n\n DQN path instead of the compact built-in model.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n FPS\/game RL observes target, HUD, movement, resource, objective, audio, and<\/p>\n\n\n\n threat signals; chooses discrete actions; records transitions; supports dry-run,<\/p>\n\n\n\n bot, assist, train, eval, and imitation workflows.<\/p>\n\n\n\n State\/action shape:<\/p>\n\n\n\n remote, and auto.<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Only use on private\/local games and servers. Start with dry-run and replay<\/p>\n\n\n\n review. Do not use it to bypass anti-cheat or automate public multiplayer play.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Strategy RL supports higher-level game decisions for League of Legends style<\/p>\n\n\n\n observations and Civilization turn-strategy decisions.<\/p>\n\n\n\n State\/action shape:<\/p>\n\n\n\n Capabilities:<\/p>\n\n\n\n victory-focus state.<\/p>\n\n\n\n Recommended use:<\/p>\n\n\n\n Use as a planning\/policy layer above game-specific perception. For Civilization,<\/p>\n\n\n\n it is better suited to turn-level advisory or private mod integration than<\/p>\n\n\n\n direct real-time control.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Driving RL is a simulation-first subsystem for lane keeping, lead following,<\/p>\n\n\n\n intersections, traffic controls, pedestrian yielding, weather\/visibility, route<\/p>\n\n\n\n progress, comfort, road-sign model response, and emergency braking.<\/p>\n\n\n\n State\/action shape:<\/p>\n\n\n\n stop, yield, speed limit, traffic-light color, warning, crosswalk, and<\/p>\n\n\n\n no-entry into the RL state vector.<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Driving-law training commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Keep this in simulation\/review. Do not wire it directly to a real vehicle<\/p>\n\n\n\n actuator. Use dry-run on and shield on by default.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Knowledge RL learns which claims to present, review, label, train, or evaluate.<\/p>\n\n\n\n It tracks claim text, epistemic labels, source trust, conflicts, confidence,<\/p>\n\n\n\n replay, and policy state.<\/p>\n\n\n\n Modes:<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Use Knowledge RL to prioritize review and labeling, not as proof. It should<\/p>\n\n\n\n surface conflicts, uncertainty, and source-trust issues before claims are used<\/p>\n\n\n\n downstream.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n This subsystem is a safety-bounded parenting coaching simulator. RL is limited<\/p>\n\n\n\n to offline simulation, shadow-mode evaluation, and parent-facing suggestions.<\/p>\n\n\n\n It is not a diagnostic or clinical decision system and never autonomously<\/p>\n\n\n\n messages, disciplines, or directs a child.<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Keep observations de-identified. Treat output as coaching suggestions for an<\/p>\n\n\n\n adult caregiver, not medical, psychological, or emergency guidance.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Political RL is a local political-news triage agent. It uses a UCB1 feed<\/p>\n\n\n\n portfolio bandit to choose RSS feeds, rewards novelty and relevance, extracts<\/p>\n\n\n\n legislative bill mentions, and flags volume spikes.<\/p>\n\n\n\n Main commands:<\/p>\n\n\n\n [text] Recommended use:<\/p>\n\n\n\n Use for local analysis and triage only. Rewards are novelty, coverage, and<\/p>\n\n\n\n relevance, not engagement or persuasion. Keep source links\/provenance attached.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Real-world RL registers RLEnv implementations that can observe real systems<\/p>\n\n\n\n and optionally actuate through host hooks. The deployment gate decides whether<\/p>\n\n\n\n any action may execute.<\/p>\n\n\n\n Built-in environment:<\/p>\n\n\n\n recommended first live RL environment because it has no physical actuation.<\/p>\n\n\n\n Hook-driven presets:<\/p>\n\n\n\n Recommended use:<\/p>\n\n\n\n Start with rl real route or llm_routing. Keep physical or operational<\/p>\n\n\n\n domains in sim, offline, or shadow until checkpoints are promoted and<\/p>\n\n\n\n action gates pass.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Differential RL connects RL to physical dynamics and control theory.<\/p>\n\n\n\n Capabilities:<\/p>\n\n\n\n Recommended use:<\/p>\n\n\n\n Use for physics simulator smoke tests, robotics-like control experiments, and<\/p>\n\n\n\n safe-RL verification before connecting policies to real actuators.<\/p>\n\n\n\n Files:<\/p>\n\n\n\n Purpose:<\/p>\n\n\n\n Distributed RL lets multiple actors push transitions to a learner, publish<\/p>\n\n\n\n parameters back to actors, run population-based training, load offline datasets,<\/p>\n\n\n\n and use LLM-based reward\/preference scoring.<\/p>\n\n\n\n Capabilities:<\/p>\n\n\n\n Recommended use:<\/p>\n\n\n\n Use when training data is generated across machines or domains. Keep the reward<\/p>\n\n\n\n rubric explicit when using RLAIF, and audit LLM-scored rewards before treating<\/p>\n\n\n\n them as training signal.<\/p>\n\n\n\n RLCurriculumOrchestrator exposes the broad catalog and safe smoke tests. Run:<\/p>\n\n\n\n [text] Current catalog IDs include:<\/p>\n\n\n\n The generic smoke domains use synthetic observations and safe action families.<\/p>\n\n\n\n They do not touch external services.<\/p>\n\n\n\n Use this sequence for any new RL domain:<\/p>\n\n\n\n Useful first commands:<\/p>\n\n\n\n [text] RL systems in this project should be treated as action-selection systems, not<\/p>\n\n\n\n general-purpose truth engines. The safe pattern is:<\/p>\n\n\n\n driving, robotics, cybersecurity, and desktop\/Windows automation.<\/p>\n\n\n\n Detailed docs:<\/p>\n\n\n\n Legacy\/reference docs:<\/p>\n\n\n\n The RL stack has a reusable C++ substrate, multiple local algorithms, MathNN DQN<\/p>\n\n\n\n support, distributed\/offline hooks, a broad curriculum catalog, and several<\/p>\n\n\n\n domain-specific systems. The highest-value operational habit is to separate<\/p>\n\n\n\n learning from actuation: train and test in simulation or offline replay, inspect<\/p>\n\n\n\n behavior, save checkpoints, evaluate, then advance through the deployment ladder<\/p>\n\n\n\n only when the policy is measurably better and safety gates pass.<\/p>\n","protected":false},"excerpt":{"rendered":" This document is the map of the reinforcement-learning surface in _AugmentedIntelligence. It explains the shared RL infrastructure, the domain modules, the training\/deployment ladder, and the commands used to inspect or run the systems. For MathNN details, see docs\/MathNN.md. For the live curriculum catalog, see docs\/RLCurriculum.md. Big Picture The RL stack has three layers: algorithms, […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-2065","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/posts\/2065","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/comments?post=2065"}],"version-history":[{"count":1,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/posts\/2065\/revisions"}],"predecessor-version":[{"id":2066,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/posts\/2065\/revisions\/2066"}],"wp:attachment":[{"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/media?parent=2065"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/categories?post=2065"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/macdaddy4sure.ai\/index.php\/wp-json\/wp\/v2\/tags?post=2065"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
<\/em>Sim -> Offline -> Shadow -> GatedLive -> Live<\/p>\n\n\n\nShared RL Infrastructure<\/h2>\n\n\n\n
System<\/strong><\/td> Main files<\/strong><\/td> Purpose<\/strong><\/td><\/tr> RL core<\/td> RLCore.hpp\/.cpp<\/td> Defines RLEnv, spaces, transitions, replay buffers, prioritized replay, RNG, and checkpoint registry.<\/td><\/tr> RL algorithms<\/td> RLAlgorithms.hpp\/.cpp<\/td> Implements tabular and lightweight value-based learners; declares remote deep-RL trainers.<\/td><\/tr> RL trainer<\/td> RLFeatures.hpp\/.cpp<\/td> Provides the unified trainer loop and umbrella rl … command dispatcher.<\/td><\/tr> Exploration and safety<\/td> RLExplore.hpp\/.cpp<\/td> Greedy, epsilon-greedy, Boltzmann, UCB, Thompson sampling, curiosity, reward shaping, and safe action shielding.<\/td><\/tr> Neural DQN<\/td> RLNeural.hpp\/.cpp<\/td> In-process DQN using MathNN online\/target Q-networks and replay.<\/td><\/tr> Agent neural registry<\/td> RLAgentNet.hpp\/.cpp<\/td> Opt-in MathNN DQN registry for named agents such as Minecraft, driving, FPS gaming, and strategy.<\/td><\/tr> Planning<\/td> RLPlanning.hpp\/.cpp<\/td> Generic MCTS\/UCT planner plus world-model interface.<\/td><\/tr> Distributed RL<\/td> RLDistributed.hpp\/.cpp<\/td> Actor\/learner transition push, parameter publishing, PBT, RLAIF scoring, and offline dataset loading.<\/td><\/tr> Differential RL<\/td> RLDifferential.hpp\/.cpp<\/td> Physics\/ODE environments, CartPole, OU exploration, stability tripwire, and reachability safety checks.<\/td><\/tr> Real-world RL<\/td> RLReal.hpp\/.cpp, RLRealEnvs.hpp\/.cpp<\/td> Real environment registry, LLM-routing bandit, and hook-driven environment presets.<\/td><\/tr> Deployment<\/td> RLDeployment.hpp\/.cpp<\/td> Sim\/offline\/shadow\/gated\/live stage gate and action-execution choke point.<\/td><\/tr> Curriculum<\/td> RLCurriculumOrchestrator.hpp\/.cpp<\/td> Catalog of RL domains and safe smoke environments.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n Algorithms<\/h2>\n\n\n\n
Local C++ Algorithms<\/h3>\n\n\n\n
\n
Remote Or Heavy Deep-RL Algorithms<\/h3>\n\n\n\n
\n
Core Commands<\/h2>\n\n\n\n
<\/em>rl status
rl seed <n>
rl registry
rl registry best <agent>
rl deep train <algo> <env> <state_dim> <action_dim> [continuous]
rl push <serialized-transition><\/p>\n\n\n\n
<\/em>rl net status
rl net enable <agent>
rl net disable <agent>
rl net save <agent> <path>
rl net load <agent> <path><\/p>\n\n\n\n
<\/em>rl real status
rl real envs
rl real route
rl real stage <agent> <sim|offline|shadow|gatedlive|live>
rl real gate <agent> <action>
rl real promote <agent> <version><\/p>\n\n\n\n
<\/em>rl curriculum status
rl curriculum plan
rl curriculum next
rl curriculum domains [category]
rl curriculum domain <id>
rl curriculum envs
rl curriculum smoke all [episodes] [max_steps]
rl curriculum smoke <domain_id> [episodes] [max_steps]
rl curriculum export [path]<\/p>\n\n\n\n
<\/em>rl domains [category]
rl domain <id>
rl sim game smoke [episodes] [max_steps]
rl sim physics smoke [episodes] [max_steps]
rl sim robot smoke [episodes] [max_steps]
rl sim energy smoke [episodes] [max_steps]
rl sim windows smoke [episodes] [max_steps]<\/p>\n\n\n\nDomain Modules<\/h2>\n\n\n\n
Minecraft DQN<\/h3>\n\n\n\n
\n
\n
<\/em>minecraft dqn status
minecraft dqn player <selector-or-name>
minecraft dqn observe
minecraft dqn train [interval_ms]
minecraft dqn eval
minecraft dqn off
minecraft dqn step
minecraft dqn dry-run
minecraft dqn qvalues
minecraft dqn explain [top_n]
minecraft dqn control server|client|both
minecraft dqn wall-guard on|off|status
minecraft dqn unstuck on|off|status
minecraft dqn interval <milliseconds>
minecraft dqn epsilon <0.0-1.0>
minecraft dqn reward <value> [terminal]
minecraft dqn save [path]
minecraft dqn load [path]
minecraft dqn net on|off|status|save <path>|load <path><\/p>\n\n\n\n
<\/em>minecraft dqn client status
minecraft dqn client focus on|off
minecraft dqn client foreground on|off
minecraft dqn client hold <milliseconds>
minecraft dqn client mine-hold <milliseconds>
minecraft dqn client use-hold <milliseconds>
minecraft dqn client mine
minecraft dqn client place
minecraft dqn client slot <1-9>
minecraft dqn client look-up
minecraft dqn client look-down
minecraft dqn client action <id>
minecraft dqn client release
minecraft dqn client title <window-title-fragment><\/p>\n\n\n\nGaming RL FPS<\/h3>\n\n\n\n
\n
\n
<\/em>gaming rl status
gaming rl features
gaming rl systems
gaming rl skills
gaming rl explain
gaming rl why
gaming rl mode off|assist|bot|train|eval
gaming rl algorithm heuristic|qtable|linear_q|sarsa|reinforce|mlp|dqn|remote|auto
gaming rl train on|off|episode start|episode end
gaming rl eval [episodes]
gaming rl selfplay on|off
gaming rl adapter set <name>
gaming rl adapter status
gaming rl perception <request>
gaming rl input hold <milliseconds>
gaming rl input mouse <max_pixels_per_tick>
gaming rl dry-run on|off
gaming rl curriculum status
gaming rl curriculum start [phase]
gaming rl curriculum next
gaming rl replay summary [path]
gaming rl replay train [path]
gaming rl bc train [epochs]
gaming rl reward good|bad|<number> [note]
gaming rl dashboard
gaming rl heatmap
gaming rl bridge spec
gaming rl checkpoint tag|list|rollback<\/p>\n\n\n\nStrategy Gaming RL<\/h3>\n\n\n\n
\n
\n
\n
\n
Driving RL<\/h3>\n\n\n\n
\n
\n
<\/em>driving rl status
driving rl features
driving rl actions
driving rl explain
driving rl dashboard
driving rl on|assist|train|eval|off
driving rl algorithm heuristic|qtable|dqn|auto
driving rl dqn on|off|status|save <path>|load <path>
driving rl scenario <name> [ticks]
driving rl curriculum status|start [phase]|next|phase <name>
driving rl adapter set <name>
driving rl adapter status
driving rl shield on|off
driving rl dry-run on|off
driving rl reward good|bad|<value>
driving rl checkpoint tag|list|rollback
driving rl save
driving rl load<\/p>\n\n\n\n
<\/em>driving rl law status
driving rl law mysql init
driving rl law seed <STATE>
driving rl law load [STATE]
driving rl law list [N]
driving rl law add <federal|state> <jurisdiction> <trigger> <required_action> <severity> <citation> <summary>
driving rl law on|off
driving rl law train [all|scenario] [ticks]
driving rl law explain<\/p>\n\n\n\nKnowledge RL<\/h3>\n\n\n\n
\n
\n
<\/em>knowledge on
knowledge off
knowledge status
knowledge claim <statement>
knowledge next
knowledge true|false|depends|unknown
knowledge label true|false|depends|unknown <claim>
knowledge conflicts
knowledge trust <source_or_claim> <0.0-1.0>
knowledge save
knowledge load
knowledge save mysql
knowledge load mysql
knowledge rl status
knowledge rl observe
knowledge rl step
knowledge rl train replay [episodes]
knowledge rl eval
knowledge rl replay [limit]
knowledge rl benchmark [episodes]
knowledge rl explain
knowledge rl use
knowledge rl stats
knowledge rl graphs [directory]<\/p>\n\n\n\nParenting \/ Child Psychology RL<\/h3>\n\n\n\n
\n
<\/em>parenting rl status
parenting rl domains
parenting rl safety
parenting rl actions
parenting rl curriculum
parenting rl observe <de-identified situation text>
parenting rl evaluate <situation text>
parenting rl train [episodes] [max_steps]
parenting rl export <path>
parenting rl reset<\/p>\n\n\n\nPolitical News RL<\/h3>\n\n\n\n
\n
<\/em>politics status
politics tick [n]
politics brief [n]
politics bills
politics save
politics load<\/p>\n\n\n\nReal-World RL Environments<\/h3>\n\n\n\n
\n
\n
\n
Differential \/ Physics RL<\/h3>\n\n\n\n
\n
\n
Distributed RL \/ RLAIF \/ Offline RL<\/h3>\n\n\n\n
\n
\n
Curriculum Domain Catalog<\/h2>\n\n\n\n
<\/em>rl curriculum domains
rl curriculum domain <id><\/p>\n\n\n\nDomain ID<\/strong><\/td> Category<\/strong><\/td> Smoke<\/strong><\/td> Purpose<\/strong><\/td><\/tr> knowledge_rl<\/td> knowledge<\/td> yes<\/td> Claim triage, critique, and memory-consolidation checks.<\/td><\/tr> gaming_rl<\/td> gaming<\/td> yes<\/td> Navigation, stuck recovery, resource timing, and reward shaping.<\/td><\/tr> game_simulator<\/td> gaming<\/td> yes<\/td> Deterministic grid navigation and obstacle handling.<\/td><\/tr> physics_simulator<\/td> simulation<\/td> yes<\/td> Force and trajectory control using local physics dynamics.<\/td><\/tr> robot_simulator<\/td> robotics<\/td> yes<\/td> Differential-drive navigation with collision and boundary penalties.<\/td><\/tr> robotics_gazebo<\/td> robotics<\/td> yes<\/td> ROS\/Gazebo navigation and manipulation staging.<\/td><\/tr> ros_robot<\/td> robotics<\/td> no<\/td> Real ROS\/hardware bridge placeholder registered at Sim.<\/td><\/tr> cybersecurity_soc<\/td> security<\/td> yes<\/td> Alert triage, evidence collection, ranking, and escalation.<\/td><\/tr> windows_ops<\/td> operations<\/td> yes<\/td> Ticket triage, patching, diagnostics, rollback planning, and audits.<\/td><\/tr> software_build<\/td> software<\/td> yes<\/td> Build, test, lint, bisect, retry, rollback, and documentation choices.<\/td><\/tr> research_science<\/td> research<\/td> yes<\/td> Literature review, hypothesis selection, experiment planning, and replication.<\/td><\/tr> healthcare_triage<\/td> healthcare<\/td> yes<\/td> Intake, guideline lookup, risk flags, and human escalation.<\/td><\/tr> education_tutoring<\/td> education<\/td> yes<\/td> Hinting, examples, retrieval practice, review, and assessment timing.<\/td><\/tr> math_algebra<\/td> mathematics<\/td> yes<\/td> Algebra step selection for MathNN DQN tutoring smoke\/training.<\/td><\/tr> math_geometry<\/td> mathematics<\/td> yes<\/td> Geometry theorem and measurement tutoring for MathNN.<\/td><\/tr> math_trigonometry<\/td> mathematics<\/td> yes<\/td> Trigonometry identity and ratio tutoring for MathNN.<\/td><\/tr> math_probability_statistics<\/td> mathematics<\/td> yes<\/td> Probability\/stats workflow tutoring for MathNN.<\/td><\/tr> math_linear_algebra<\/td> mathematics<\/td> yes<\/td> Linear algebra matrix\/system tutoring for MathNN.<\/td><\/tr> math_finite_math<\/td> mathematics<\/td> yes<\/td> Finite math (sets, combinatorics, LP, graphs) for MathNN.<\/td><\/tr> child_parenting<\/td> parenting<\/td> yes<\/td> Routines, de-escalation, boundaries, choices, and safety escalation.<\/td><\/tr> finance_portfolio<\/td> finance<\/td> yes<\/td> Allocation, risk balancing, rebalance timing, and drawdown-aware simulation.<\/td><\/tr> stock_research<\/td> finance<\/td> yes<\/td> RSS reading, sentiment checks, fundamentals review, and watchlist updates.<\/td><\/tr> product_recommendation<\/td> commerce<\/td> yes<\/td> Relevance, diversity, safety filtering, preference questions, and audits.<\/td><\/tr> sales_pricing<\/td> commerce<\/td> yes<\/td> Offer timing, guarded discounts, follow-up cadence, and churn handling.<\/td><\/tr> supply_chain<\/td> operations<\/td> yes<\/td> Reorder, supplier, routing, buffer stock, and disruption response.<\/td><\/tr> energy_grid<\/td> infrastructure<\/td> yes<\/td> Load forecasting, storage dispatch, outage response, and demand shifting.<\/td><\/tr> legal_research<\/td> legal<\/td> yes<\/td> Statute\/case lookup, issue spotting, citation checks, and attorney escalation.<\/td><\/tr> civic_policy<\/td> government<\/td> yes<\/td> Stakeholder tradeoffs, sequencing, equity checks, review, and audits.<\/td><\/tr> manufacturing_quality<\/td> manufacturing<\/td> yes<\/td> Inspection, calibration, sampling, rework, and line-stop recommendations.<\/td><\/tr> media_workflow<\/td> creative<\/td> yes<\/td> Ingest, tagging, edit queues, rendering, captioning, and publishing review.<\/td><\/tr> personal_productivity<\/td> personal<\/td> yes<\/td> Prioritization, scheduling, focus blocks, breaks, reminders, and review.<\/td><\/tr> smart_home<\/td> iot<\/td> yes<\/td> Comfort, energy, notifications, fallback-safe automation, and audit policy.<\/td><\/tr> fleet_maintenance<\/td> operations<\/td> yes<\/td> Inspection, preventive service, parts staging, rerouting, and downtime reduction.<\/td><\/tr> call_center<\/td> customer_ops<\/td> yes<\/td> Routing, response suggestions, clarifying questions, escalation, and QA.<\/td><\/tr> llm_routing<\/td> llm<\/td> no<\/td> Backend selection by latency, quality, cost, and task fit.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n Training And Evaluation Workflow<\/h2>\n\n\n\n
\n
<\/em>rl curriculum smoke all 8 96
rl curriculum next
rl net status
rl real envs<\/p>\n\n\n\nSafety Model<\/h2>\n\n\n\n
\n
Documentation Map<\/h2>\n\n\n\n
\n
\n
Short Summary<\/h2>\n\n\n\n