← Adaptogenic Wellness Platform

How It All Connects

A guide to the kask Adaptogen Platform — how the data domains interrelate, how to read the evidence scores, how to use each tool, and the principles the adaptogen_curator agent follows when answering your questions.

Contents
  1. The Nine Data Domains
  2. How Data Flows Through the System
  3. Reading the Evidence Scores
  4. The Safety-First Principle
  5. Traditional Medicine Meets Modern Science
  6. Using the Tools
  7. Building a Tonic — Step by Step
  8. Building a Serum — Skin Microbiome
  9. The Gustatory Cortex Simulator
  10. How the Agent Uses This Data
  11. Limitations & Honest Gaps

1. The Nine Data Domains

The database is organised into nine domains, all anchored by the SPECIES table. Every species carries full Linnaean classification, external database identifiers (NCBI, GBIF, IPNI), IUCN conservation status, and native range. Everything else connects back to a species.

1

Taxonomy & Identity

The anchor. Scientific names, common names in 15+ languages, images with license metadata, and adaptogen classification (primary vs secondary, stress axes modulated).

species · species_image · adaptogen_profile
2

Traditional Medicine

Nine medicine systems (TCM, Ayurveda, Kampo, Unani, Siddha, Sowa Rigpa, Jamu, Korean, Western), 20 classical texts spanning 3000 years, and structured traditional use records with classical text passage references.

medicine_system · classical_text · traditional_use · traditional_use_text_ref
3

Phytochemistry

169 bioactive compounds with PubChem CIDs, SMILES, molecular weights, compound classes, and biosynthetic pathway annotations. Linked to species with plant part, extraction method, and concentration data.

compound · species_compound
4

Evidence & Research

88 indexed studies, 93 evidence ratings with triple-framework scoring, 26 systematic reviews, and 21 registered clinical trials. All PubMed IDs and NCT numbers are real references.

indication · study · evidence_rating · systematic_review · clinical_trial
5

Safety & Interactions

48 pharmaceutical drugs, 97 herb-drug interactions (with severity, mechanism, and management), condition contraindications, adverse events, and CYP450 enzyme interactions tied to specific compounds.

drug · herb_drug_interaction · condition_contraindication · adverse_event · cytochrome_interaction
6

Genetics & Genomics

Authentication markers (ITS2, rbcL, matK) for adulteration detection, genome assemblies, annotated genes with biosynthetic pathway relevance, and population/chemotype variants affecting bioactive yield.

genomic_marker · genome_assembly · gene · population_variant · external_db_reference
7

Indications & Conditions

49 medical indications with ICD-11 and MeSH cross-references, organised into 14 therapeutic categories from psychiatric through oncological. Each carries a detailed clinical description.

indication (with icd11_code, mesh_term, category)
8

Sensory Receptor Pharmacology

Taste receptors (TAS1R/TAS2R), olfactory receptors, trigeminal channels (TRPV1/TRPM8), binding affinities, sensory descriptors, mixture interactions, and the bridge from TCM five-taste theory to modern receptor science.

receptor · compound_receptor_binding · olfactory_receptor_binding · trigeminal_activity · taste_descriptor · compound_taste_percept · tcm_taste_receptor_map · mixture_interaction
9

Gustatory Cortex Simulation

Neural population models, simulation configs, input vectors (compound→receptor activation), GC response profiles with 128-dim taste embeddings, empirical validation, and limbic projections predicting hedonic valence and HPA axis implications.

gc_neuron_population · gc_simulation_config · gc_simulation_run · gc_simulation_input · gc_empirical_validation · gc_simulation_run_validation · limbic_projection

2. How Data Flows Through the System

The system has three main pipelines. Each one answers a different question.

Pipeline A — "What should I take?"

Desired Outcome (indication) → Evidence Ratings (species ranked by star/GRADE/Oxford) → Safety Check (drug interactions for your medications) → Compound Profile (what bioactives, at what concentration) → Traditional Preparation (how to prepare, what dose) → Tonic Formula (personalised blend)

Pipeline B — "Is it safe with my medication?"

Your Medication (drug table) → Herb-Drug Interaction (severity, mechanism) → CYP450 Interaction (which enzyme, which compound drives it) → Management Recommendation (avoid / monitor / adjust) → Contraindications (pregnancy, lactation, conditions)

Pipeline C — "What will it taste like?"

Species + Compounds + Doses → Compound Classes → Heuristic Flavor Profile → TCM Taste Data (五味 nature/taste/meridian) → Flavor Radar (10-axis spider chart) → Receptor Activation Vector (Domain 8) → Gustatory Cortex Simulation (Domain 9) → Hedonic Score + Predicted TCM Taste → Limbic Projection → HPA Axis Implication
Key insight Pipeline C is where traditional medicine and modern neuroscience converge. The TCM five-taste classification (甘苦酸咸辛) maps to specific taste receptor activations, which produce specific gustatory cortex response patterns, which project to limbic structures that modulate the HPA axis — the very axis that defines adaptogenic activity. The taste of a herb may not be incidental to its therapeutic action; it may be the signal for it.

3. Reading the Evidence Scores

Every species–indication pair is scored on three frameworks simultaneously, serving three different audiences:

Framework Audience Scale What It Measures
GRADE Researchers, clinicians 1–4 (very low → high) Certainty in the estimate of effect — how likely is it that further research will change the conclusion?
Oxford CEBM Evidence-based medicine 1–5 (SR of RCTs → expert opinion) Study design hierarchy — what type of evidence supports the claim?
Star ★ Consumers, developers 1–5 stars Pragmatic confidence — "how sure should I be that this works?"

How to interpret star ratings

Stars Meaning Example
★★★★★ Strong evidence from multiple high-quality RCTs; widely accepted (No adaptogen currently reaches 5 stars — reserved for pharmaceutical-grade evidence)
★★★★ Good evidence from RCTs and/or strong systematic reviews Ashwagandha for stress/anxiety, Kava for GAD, Boswellia for OA, Black Seed for diabetes
★★★ Moderate evidence; some RCTs but with limitations Rhodiola for burnout, Lion's Mane for MCI, Bacopa for memory, Tulsi multi-domain
★★ Preliminary — mostly animal/in-vitro or very small trials Chaga (extensive preclinical, zero RCTs), Ashitaba (Nature Medicine preclinical), Maca for menopause
★ Traditional use only; no meaningful clinical data (Rarely assigned — most species in the database have at least some clinical research)

Additional effect characterisation

Each evidence rating also captures:

  • Effect direction — positive, negative, neutral, or mixed
  • Effect size — large, moderate, small, negligible, or not estimable
  • Clinical significance — clinically significant, statistically significant only, not significant, or insufficient data
  • Hybrid summary — a plain-language synthesis of the evidence for that species-indication pair
  • Reviewer notes — scholarly commentary on the evidence quality, limitations, and context

4. The Safety-First Principle

Core rule Every therapeutic claim is presented alongside its safety profile. The agent will never recommend a species without surfacing its herb-drug interactions, contraindications, and adverse events. Benefits and risks are always co-presented.

Interaction severity levels

Severity Count Meaning Action Required
Major 17 Life-threatening or requiring medical intervention. Transplant rejection, hepatotoxicity, serotonin syndrome, hypertensive crisis. >AVOID the combination. No safe dose exists.
Moderate 37 Clinically significant — requires monitoring, dose adjustment, or clinical judgement. MONITOR — use with caution under medical supervision. Adjust doses, check labs.
Minor 18 Unlikely to require intervention at standard doses. Theoretical or additive effects. AWARENESS — inform your healthcare provider. Usually tolerable.

Special populations

The database flags three high-priority special populations for every species with contraindication data:

  • Pregnancy — many adaptogens that move blood (Dong Quai), stimulate hormonal axes (Ashwagandha HPT), or contain emmenagogues are contraindicated
  • Lactation — some adaptogens pass into breast milk; Shatavari is specifically a galactagogue (promotes milk)
  • Paediatric — dose-response differs in children; many adaptogens lack paediatric safety data

The He Shou Wu lesson

Reynoutria multiflora (He Shou Wu / Fo-Ti) is the clearest illustration of why safety data matters. The raw form contains emodin (hepatotoxic anthraquinone) and has caused 450+ documented cases of liver injury. The processed form (nine-steaming, nine-drying with black bean juice) reduces emodin by 85–95% while concentrating the tonic stilbene TSG. Same species, two radically different safety profiles. The database tracks both forms with separate compound concentrations.

5. Traditional Medicine Meets Modern Science

Principle Traditional medicine systems are respected as knowledge systems — not reduced to pharmacological mechanisms. TCM, Ayurveda, Kampo, and other traditions represent millennia of empirical clinical observation. Modern pharmacology provides mechanistic explanations, not replacements.

How TCM classification maps to modern data

  • → TCM Nature (四气) warm/cool/cold/hot/neutral maps to thermogenic and vasodilatory compound profiles
  • → TCM Taste (五味) sweet/bitter/sour/salty/pungent maps to specific taste receptor activations (TAS1R, TAS2R, OTOP1, ENaC, TRPV1) via the tcm_taste_receptor_map table
  • → TCM Meridian/Organ (归经) "enters Liver and Kidney" potentially maps to the limbic projection patterns from gustatory cortex to hypothalamus and autonomic centres
  • → TCM Actions (功效) "tonifies qi" / "invigorates blood" map to measurable pharmacological endpoints (immunomodulation, haematopoiesis, cortisol modulation)

The Ayurvedic Rasayana concept

Rasayana (rejuvenation) in Ayurveda closely parallels the modern adaptogen concept. Both describe substances that enhance resilience, promote longevity, and improve resistance to stress without causing harm. The Charaka Samhita's criteria for a Rasayana herb overlap remarkably with Brekhman's 1968 adaptogen criteria — formulated independently on opposite sides of the world, 2000 years apart.

The four Medhya Rasayanas

Charaka Samhita names exactly four herbs as Medhya Rasayana (intellect-rejuvenators): Brahmi (Bacopa monnieri), Mandookaparni (Centella asiatica), Yashtimadhu (Glycyrrhiza glabra), and Shankhapushpi (Convolvulus pluricaulis). Three of these four are in our database with clinical evidence for cognitive enhancement — a 3000-year-old classification validated by 21st-century RCTs.

6. Using the Tools

◎
>Species Browser

Browse all 60 species with filterable grid. Click any row to expand a detail panel showing adaptogen profile, images, compounds, evidence ratings (with star/GRADE scores), traditional medicine data (TCM nature/taste/meridian, Ayurvedic classification), and safety profile (drug interactions sorted by severity).

Use when: you want to explore a specific species in depth, or compare species within a category (e.g. "show me all primary adaptogens" or "filter by Lamiaceae family").
◇

Compound Pathway Mapper

Select desired health outcomes, then see a Sankey diagram mapping the flow from conditions through pharmacological actions through compounds to species. Link thickness reflects evidence strength. Red borders flag species with major drug interaction alerts. Click any node for detail. Table view available.

Use when: you want to understand why certain species help with certain conditions — what compounds are responsible, what actions they perform, and which species contain them.
▦

Evidence Heatmap

A matrix of species (rows) × indications (columns), colored by star rating. Instantly see which species have the broadest evidence base, which conditions have the most options, and where the evidence gaps are. Click column headers to sort. Hover cells for rich tooltips with GRADE, effect direction, and summaries.

Use when: you want a bird's-eye view of the entire evidence landscape — "what has the most evidence for what?" — or when comparing multiple species for the same condition.
⚗

Tonic Builder

Design personalised adaptogenic formulas. Select outcomes, enter your medications, and the system suggests ingredients ranked by evidence match (penalising species with drug conflicts). Adjust doses with sliders. See the predicted flavor profile (10-axis radar chart), GC response preview, safety panel, and evidence trail — all updating in real-time. Eight preset formulas available.

Use when: you want to design an actual tonic or supplement formula and need to balance therapeutic goals, safety constraints, and palatability simultaneously.
✦

Serum Maker

Design personalised botanical serums for the skin microbiome. Select skin concerns (acne, barrier damage, ageing, hyperpigmentation, rosacea), enter your skin type and medications, and the system suggests a formula with predicted skin property profile (8-axis radar: hydrating, anti-inflammatory, antimicrobial, antioxidant, barrier repair, collagen boost, brightening, soothing), penetration & delivery profile (surface → epidermis → dermis → microbiome), microbiome impact assessment, INCI preview, safety checks including photosensitivity and sensitisation, and evidence trail. Six preset serums available.

Use when: you want to formulate a topical botanical serum targeting specific skin concerns — the dermatological counterpart to the Tonic Builder, with skin-type-aware scoring and microbiome-conscious ingredient selection.

7. Building a Tonic — Step by Step

  1. Define your goals. Select 1–3 desired outcomes from the chips (e.g. "Anxiety & Stress Relief" + "Cognitive Enhancement"). Fewer, more focused goals yield better suggestions than selecting everything.
  2. Enter your medications. Type any prescription or OTC medications into the medication input. The system will automatically penalise species with major drug conflicts and flag interactions in the safety panel. This step is critical for personalisation.
  3. Review the auto-suggested formula. The system ranks all 60 species by evidence match to your selected outcomes, penalises those conflicting with your medications, and suggests the top 5. Review the ingredient list — each shows the species image, common name, family, and top 3 pharmacological actions.
  4. Adjust ingredients. Remove species that don't fit your goals or taste preferences. Add others using the search. Adjust doses with the sliders (0.5–10g per ingredient). Watch the flavor radar and GC response update in real-time.
  5. Check the flavor profile. The 10-axis radar chart shows predicted taste dimensions (Sweet 甘, Bitter 苦, Sour 酸, Salty 咸, Pungent 辛, Astringent 涩, Aromatic, Earthy, Cooling, Warming). If the formula is too bitter, add a sweet balancer (liquorice, cinnamon). If too earthy, add an aromatic (lemon balm, chamomile).
  6. Check the hedonic score. The GC response strip shows a hedonic valence bar from "Pleasant" to "Intensely Bitter." A formula you won't drink is a formula that won't work. Target "Mild" or better.
  7. Review safety. The safety panel shows all interactions between your formula and your medications. Major alerts (red) mean do not combine. Moderate alerts (orange) mean monitor. Any remaining major alerts should be resolved by removing the conflicting ingredient.
  8. Review the evidence trail. Each ingredient-outcome pair shows its star rating and GRADE certainty. A formula with ★★★★ ingredients for your selected outcomes is well-supported. A formula with ★★ ingredients is more speculative.
Pro tip — TCM harmonisation Classical TCM formulae always balance herbs across roles: a jun (emperor/chief) herb that targets the primary condition, a chen (minister) that supports it, a zuo (assistant) that moderates side effects, and a shi (envoy) that directs the formula to target organs. You can apply this logic: pick your primary adaptogen for the main goal, add a supporting adaptogen, include a safety-balancing herb (e.g. liquorice as harmoniser), and add a flavor-directing herb (e.g. ginger as warming envoy, mint as cooling envoy).

7b. Building a Serum — Skin Microbiome

The Serum Maker applies the same evidence-driven formulation logic as the Tonic Builder, but adapted for topical skin applications. Instead of flavor profiles and gustatory cortex simulation, it models skin penetration, microbiome impact, and dermatological safety.

  1. Select your skin type. Choose from normal, dry, oily, combination, sensitive, or mature. This adjusts ingredient scoring — sensitive skin boosts anti-inflammatory species, oily skin boosts antimicrobials, mature skin boosts anti-ageing compounds.
  2. Select skin concerns. Pick 1–3 concerns from categories like Acne & Oil, Redness & Sensitivity, Barrier & Hydration, Ageing & Firmness, Tone & Radiance, or Repair. The system maps each concern to relevant pharmacological actions and scores species accordingly.
  3. Enter medications. Same drug interaction checking as the Tonic Builder — topical botanicals can still interact with systemic medications (e.g. anticoagulants, immunosuppressants).
  4. Review the formula. Adjust concentrations with sliders (0.5–5%). Typical botanical serum actives range 1–3%. The system caps at 6 botanicals to keep formulations practical.
  5. Read the skin property radar. The 8-axis chart shows your formula's predicted profile across hydrating, anti-inflammatory, antimicrobial, antioxidant, barrier repair, collagen boost, brightening, and soothing dimensions. Balance according to your concerns.
  6. Check penetration profile. Shows predicted delivery across four layers: surface (where polysaccharides and large molecules act), epidermis (flavonoids, phenolics), dermis (small lipophilic terpenoids, alkaloids), and microbiome (compounds that interact with skin flora).
  7. Review microbiome impact. Shows how your formula may affect the skin microbiome — selective antimicrobial activity, prebiotic potential from polysaccharides, immunomodulatory effects on AMP production, and anti-inflammatory reduction of dysbiosis-promoting signals.
  8. Check the INCI preview. Auto-generated INCI (International Nomenclature of Cosmetic Ingredients) list in descending concentration order — useful for label compliance and communicating formulations to contract manufacturers.
Key difference from the Tonic Builder Topical formulations interact with the skin microbiome — the ~1000 species of bacteria, fungi, and viruses that colonise the skin surface and play critical roles in barrier function, immune education, and pathogen defence. The Serum Maker considers this ecology: beta-glucan polysaccharides act as prebiotics for beneficial S. epidermidis, while selective antimicrobials (thymoquinone, eugenol, EGCG) can reduce pathogenic C. acnes and S. aureus without sterilising the entire community. This microbiome-aware approach distinguishes botanical serums from conventional antiseptic formulations.

8. The Gustatory Cortex Simulator

The ultimate goal of Domains 8 and 9 is to predict how the brain processes the taste of an adaptogenic compound mixture. This is not yet implemented as a running simulation — the database tables (migrations 007 and 008) define the infrastructure for future development.

The pipeline (planned)

Compound concentrations in the formula × Binding affinities from COMPOUND_RECEPTOR_BINDING = Receptor activation vector (how strongly each taste receptor fires) Receptor activation vector → Simulated taste receptor cell responses → Nucleus tractus solitarius relay → Gustatory cortex (anterior insula) — taste IDENTITY → Gustatory cortex (posterior insula) — taste HEDONIC VALUE → Limbic projections (amygdala, OFC, hypothalamus) → Predicted TCM taste category (closing the loop) → HPA axis implication (the adaptogen connection)

Why this matters for tonics

Currently, the Tonic Builder's flavor profile is heuristic — compound class → taste mapping based on known sensory properties. When the receptor binding data is populated (Domain 8) and the GC simulation is implemented (Domain 9), the flavor prediction will become pharmacologically grounded: computed from actual receptor activation magnitudes determined by compound concentrations and binding affinities. This would enable:

  • Predicting the taste of a novel compound mixture before it's prepared
  • Optimising formulas for palatability by adjusting ratios to shift receptor activation patterns
  • Understanding why certain traditional combinations taste the way they do (and whether that taste predicts efficacy)
  • Generating 128-dimensional "taste fingerprints" for similarity search ("find formulas that taste like this one")

9. How the Agent Uses This Data

Agent identity The adaptogen_curator is a compound research agent in the Agent Bestiary. It holds the complete adaptogen database as its seed ontology and registers with Xaman Ek (the platform navigator). It exposes 12 domain-specific MCP tools and composes with other agents (genome_profiler, macro_forecaster, entity_investigator, naturalist, observation_analyst, coherence_evaluator) for research tasks.

When you ask a question, the agent follows this logic:

  1. Identify the query type: Is it a species query? A safety query? An evidence query? A formulation question? A traditional medicine question? A compound/pathway question?
  2. Query the relevant domain(s): A simple species question touches Domain 1. A safety question touches Domains 1 + 5. A tonic formulation question may touch all nine domains.
  3. Apply the safety-first principle: Before presenting any therapeutic claim, check for major drug interactions, contraindications, and adverse events. Surface these alongside the benefits.
  4. Cite evidence tiers: Every claim is tagged with its GRADE certainty and star rating. The agent distinguishes "meta-analysis of 14 RCTs" from "single animal study" in its language.
  5. Respect traditional knowledge: When discussing TCM, Ayurvedic, or other traditional uses, the agent presents them as knowledge systems with their own internal logic — not as "folk remedies awaiting scientific validation."
  6. Compose with other agents when needed: For genomic questions, it delegates to genome_profiler. For market analysis of adaptogenic products, it calls macro_forecaster. For research network mapping, it uses entity_investigator.

Example agent interactions

  • "What is the evidence for ashwagandha in anxiety?" → Evidence query (Domain 4), with safety context (Domain 5) and traditional use context (Domain 2)
  • "Compare rhodiola and ginseng for chronic fatigue" → Multi-species comparison across evidence (Domain 4), compounds (Domain 3), and actions (Domain 1)
  • "I take warfarin — which adaptogens should I avoid?" → Safety query (Domain 5) filtered by drug, returning all species with warfarin interactions sorted by severity
  • "What does the Charaka Samhita say about ashwagandha?" → Traditional medicine query (Domain 2), returning text references with original passage and translation
  • "Design a sleep formula that avoids blood-sugar-lowering herbs" → Formulation query across all domains with constraint filtering

10. Limitations & Honest Gaps

This platform is a research and exploration tool, not a medical device. Understanding its limitations is as important as using its features.

Evidence gaps

  • No adaptogen currently reaches ★★★★★. Even the best-studied species (ashwagandha, ginseng, curcumin) have evidence limitations — small studies, heterogeneous preparations, limited replication across independent research groups.
  • Many species have more traditional evidence than clinical evidence. Chaga has centuries of Russian/Finnish folk use and a registered pharmaceutical (Befungin) but zero published RCTs. The evidence heatmap makes these gaps visible.
  • Preparation matters enormously. An ethanol extract is pharmacologically different from a water decoction of the same species. Most clinical trials use standardised extracts that may not match commercially available products.

Flavor prediction limitations

  • The current flavor radar is heuristic, not pharmacological. It maps compound classes to taste profiles based on known sensory properties, TCM taste classifications, and species-specific overrides. It does NOT compute receptor activation from binding affinities (that's the future GC simulation pipeline).
  • Individual taste perception varies significantly based on genetics (TAS2R38 polymorphism affects bitter perception by 10–100×), age, medications, and health status. The predicted flavor is a population average, not a personal prediction.

Safety limitations

  • The 97 documented herb-drug interactions represent the known interactions. Absence of an interaction in the database does not mean absence of interaction in reality. Many adaptogen-drug combinations have simply never been studied.
  • Individual responses to herbs vary. Idiosyncratic reactions (like He Shou Wu hepatotoxicity in susceptible individuals) are unpredictable from population-level data.
  • This tool does not replace medical advice. Always consult a healthcare professional before combining adaptogenic herbs with prescription medications, during pregnancy/lactation, or for serious health conditions.

Data freshness

  • The current dataset is a curated snapshot. New studies, trials, and safety reports are published continuously. The Phase 3 data pipeline (PubMed, ClinicalTrials.gov, PubChem automated ingestion) is planned but not yet implemented.
  • All PubMed IDs, DOIs, NCT numbers, and NCBI/GBIF identifiers in the database are real and can be independently verified. Nothing is fabricated.