Domain MCP Server
Thailand Relocation MCP
Visa pathways, budgeting, area & schooling intelligence, rentals & acquisition, healthcare, culture adaptation - exposed as typed tools, stable resources & guided prompt templates. Designed for adaptive, profile-aware planning.
- ▹Schema-first relocation knowledge surface
- ▹Deterministic prompt frames
- ▹Education + Real Estate fusion
- ▹Composable reasoning chains
Tools
20+
Prompts
10+
Resources
40+
Quick Navigation
Optimized for inspection & multi-channel reuse (web + chat + agents).
This surface is an exploratory domain MCP case study – a fully functioning prototype demonstrating how relocation knowledge can be normalized into tools, stable resources & deterministic prompt contracts. It is not a commercial product or paid service. The focus is pattern validation: schema design, chaining reliability, multi-channel reuse and cross-domain fusion (visa ⇄ budget ⇄ area ⇄ education ⇄ property) without prompt bloat.
If you want to adapt the pattern (new country, multi-country layer, internal planning console) or collaborate on open evolution, reach out with scope & context and I can outline an incremental path.
From MCP Server to Platform Surface
This relocation MCP underpins the broader platform experience (public assistant, Telegram bot and future internal planning console). A single protocol surface feeds every channel - no duplicated logic or content drift. Phad is the conversational layer sitting on top of the same tool + prompt + resource substrate.
- ▹Unified data manager → multi-channel outputs
- ▹Phad orchestration = prompt + tool chaining UX
- ▹Deterministic templates enable regression baselines
- ▹Selective resource URIs prevent token bloat
- ▹Future: multi-country expansion by swapping dataset mounts
Phad Assistant
Lightweight conversational shell orchestrating schema-first planning flows across visa, budget, housing & schooling domains.
Phad reuses the exact same tool & prompt contracts - its differentiation is sequencing + channel UX, not hidden prompt bundles.
Access / Integration
Trial / integration access available (SSE transport). Share use case + intended channel (web, chat, internal agent) via contact. Scoped credentials or limited prompt + tool set can be provisioned for evaluation.
Request EvaluationCurrent Deployment
Single MCP surface powers public multilingual assistant + Telegram bot with consistent visa, budget, housing, healthcare & schooling reasoning. No duplicated logic across channels.
- ▹Public assistant (web)
- ▹Telegram bot trials
- ▹Internal planning briefs
Integrated Real Estate Layer
Full property intelligence set (search, comparable sizing, cost estimation, ownership structure, due diligence) merged into relocation data manager. Enables cross-domain chaining (visa → budget → area → property) without token inflation.
- •Closing cost estimator
- •Ownership structure guidance
- •Due diligence checklist
- •Comparable sizing + area cost multipliers
- •Agent matching scaffolds
- •Chain into rental / budgeting flows
New Education Domain
Structured schooling + alternative education dataset: curricula, phases, fee bands, SEN support & early childhood models. Exposed via stable resources (e.g. thailand://education/schools) + modular tools for selective prompt fusion.
🏫Schools
International & bilingual catalog (curriculum, fee bands, strengths).
🧩Alternative
Homeschooling & hybrid models with regulatory notes.
👶Early Childhood
Nursery & pedagogy variants (Montessori, Reggio) + fee ranges.
🎓Higher Ed
Bridge programs & pathway sequencing.
🧠SEN Support
Service taxonomy + support center indexing.
🗂️Composite
School landscape summarization + shortlist reasoning.
Surface Snapshot
- ▹ 20+ tools (visa, budgeting, rental, real estate, education, culture, healthcare)
- ▹ 10+ prompt templates (visa strategy, budget scenarios, composite relocation, due diligence)
- ▹ Stable resources:
thailand://visa/types,thailand://areas,thailand://education/schools,thailand://realestate/ownership - ▹ Parameterized templates:
thailand://budget/estimate/{city}/{household}/{level}
Tool Catalog
Identifiers only; logic summarized. All return structured JSON for deterministic chaining.
🛂Visa & Legal
- •get_visa_types
- •list_visa_routes
- •get_visa_route
- •get_visa_checklist
Structured route + checklist modeling.
💸Budgeting
- •estimate_budget
- •compare_cities_multi
- •get_city_multipliers
Ratio-based cost modeling.
Prompt Catalog
Deterministic, pure templates that orchestrate tool output into analytical frames (no hidden side-effects).
🧪Visa Strategy
Route matrices & risk framing
- ▹visa_strategy_overview
- ▹visa_route_comparison
🧪Composite Planning
Holistic relocation sequencing
- ▹relocation_composite_plan
Prompt Architecture & Design Philosophy
🔁Pure & Deterministic
No network / time variance; reproducible diffs.
📑Section Contracts
Stable scaffolds: header → extracts → analysis → actions.
🧪Selective Fusion
Curated subsets; avoids token glut yet preserves anchors.
🧷Normalized Vocabulary
Synonyms collapsed earlier for precision follow-ups.
📊Scoring Hints
Explicit bands; internal weighting remains private.
🧵Chainable & Idempotent
Repeated calls stable; enrichment appends not rewrites.
Example Plan Pattern
Simplified relocation composite structure (abbreviated).
Representative Flows
Query → tool + prompt chaining patterns (external contract only).
Budget + Areas
Query: Solo remote worker budget & shortlist
- estimate_budget
- get_area_cost_snapshot
- budget_scenario_analysis
- area_comparison_scorecard
Shape: Budget table + ranked scorecard
Example Calls
Illustrative JSON-RPC calls (truncated).
Design Highlights
- •Immutable JSON load once
- •Domain module separation
- •Pure prompts (no I/O)
- •Structured calculators embedded
- •Deterministic chaining surface
- •Low-drift evolution path
Emerging Pattern
Domain-normalized data + semantic tools + guided prompts transform relocation help from generic lists into adaptive, profile-aware planning sequences.
Road Ahead
Housing developer catalogs, richer area scoring, multi-city scenario planner - without breaking existing contracts.
🧭Developer Catalogs
PlannedProject-level listing intelligence.
- ▹Project metadata
- ▹Developer profiles
- ▹Risk signals
🛰️Multi-City Planner
ExploringScenario deltas across cities.
- ▹Cost variance
- ▹Lifestyle differentials
- ▹Visa constraints
Want a domain-specific relocation MCP (Vietnam, Malaysia, multi-country) or to adapt this pattern? Share scope & maturity - I'll outline an incremental sequencing plan.
Start a Conversation