io.github.SamYatesSmith/tru8 — Model Context Protocol (MCP) Server

This MCP server, “Tru8,” provides source-traced evidence research for AI agents. It organizes evidence retrieved from the open web and specialist APIs, classifies evidence by tier and type, decomposes claims into elements, maps evidence to those elements, and presents a structured evidence landscape for review.

🛠️ Key Features

Organizes source-traced evidence for AI agents
Evidence classification by tier (primary / reporting / commentary)
Evidence classification by type (data / official / news / analysis / opinion / academic)
Decomposes claims into elements and maps evidence to elements
Retrieves evidence from the open web and specialist APIs
Structured presentation of the full evidence landscape (not summarised)

🚀 Use Cases

Claim or URL submission for evidence retrieval
Claim verification and fact-checking workflows
Evidence organization for LLM tools and agent decision-making

⚡ Developer Benefits

Clear evidence-to-claim-element mapping
Consistent tier/type labels to guide downstream processing
Supports MCP-based integration for LLM tools

⚠️ Limitations

Tool availability is limited to 3 tools (toolCount: 3)
Requires Docker & Docker Compose, Python 3.11+, and Node.js 20+

Tru8 — AI-Powered Evidence Research Platform

Tru8 organises evidence so you can decide what matters. Submit a claim or URL, and Tru8 retrieves evidence from the open web and specialist APIs, classifies it by tier (primary / reporting / commentary) and type (data / official / news / analysis / opinion / academic), decomposes claims into elements, maps evidence to those elements, and presents the full landscape — structured, not summarised.

Mission: "We organise; you decide."

Quick Start

Prerequisites

Docker & Docker Compose
Python 3.11+
Node.js 20+

1. Start Infrastructure

bash

docker-compose up -d    # PostgreSQL 16, Redis 7, Qdrant

2. Backend

bash

cd backend
python -m venv venv
source venv/bin/activate    # Windows: venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env        # Edit with your API keys
alembic upgrade head
uvicorn app.main:app --reload    # http://localhost:8000

3. Frontend

bash

cd web
npm install
npm run dev    # http://localhost:3000

Project Structure

code

tru8/
├── backend/              # FastAPI backend + evidence pipeline
│   ├── app/
│   │   ├── api/v1/       # REST endpoints (checks, agent, verify, payments)
│   │   ├── core/         # Config, auth, pricing, manifest signing
│   │   ├── models/       # SQLModel schemas (Check, Claim, Evidence, ClaimConsensus)
│   │   ├── pipeline/     # Two-phase evidence pipeline
│   │   ├── services/     # API adapters, search, payments, consensus
│   │   └── utils/        # Classifiers, deduplication, query planning
│   ├── alembic/          # Database migrations
│   ├── tru8_mcp/         # MCP server package for AI agents
│   └── tests/            # 990 unit + integration tests
├── web/                  # Next.js 14 frontend
│   ├── app/              # Pages (dashboard, developers, public reports)
│   ├── components/       # Evidence views, claim map, marketing
│   └── lib/              # API client, tiers, diagnostics
├── shared/               # Shared TypeScript types
├── docker-compose.yml    # PostgreSQL, Redis, Qdrant, MinIO (optional)
└── .claude/CLAUDE.md     # Full engineering context

Pipeline

Two-phase pipeline with user claim selection gate (article mode):

code

Phase 1:  Ingest → Extract claims → Classify article → Rank claims
          [PAUSE — user selects claims]

Phase 2:  Fact-check lookup → Decompose into elements → Retrieve evidence →
          Score relevance → Classify tier/type → Map evidence to elements →
          Derive orientation → Complete

Post:     Coverage recovery (Stage 5.1), video recommendations, auto-archiving

Evidence flows through web search plus specialist source adapters:

Category	Sources
Web search	Serper.dev, Brave Search, SerpAPI (fallback chain)
Fact-check	Google Fact-Check API
Government	GOV.UK, Hansard, GovInfo, Companies House, ONS
Legal	UK Legislation (legislation.gov.uk)
Economic	FRED, Marketaux, World Bank
Academic	Semantic Scholar, OpenAlex, PubMed
Climate	NOAA, WeatherAPI, Open-Meteo, GBIF
Archives	Wikipedia (+ reference mining), Library of Congress, Internet Archive
Health	WHO
Sports	Transfermarkt, Football-Data.org
Video	YouTube Data API

Six Evidence Views

View	Question it answers
Cartographer	What's the shape of this conversation?
Librarian	Is this the full set, clearly labelled?
Interpreter	Does this evidence answer this sub-question?
Projectionist	What's been said on camera?
Chronologist	When did the evidence appear?
Seeker	What don't we know yet?

API & Agent Access

Tru8 exposes a developer API and MCP server for AI agents.

Agent Commerce Gateway

bash

# Quick check (~15s, heuristic classification)
curl -X POST https://api.trueight.com/api/v1/agent/quick \
  -H "X-API-Key: $TRU8_API_KEY" \
  -d '{"claim": "UK inflation is 3.2%"}'

# Smart endpoint with tier fallback (lookup → consensus → quick → full)
curl -X POST https://api.trueight.com/api/v1/agent/check \
  -H "X-API-Key: $TRU8_API_KEY" \
  -d '{"claim": "UK inflation is 3.2%", "max_tier": "full"}'

Tier	Cost	Time	What you get
Lookup	~£0.02	instant	Cached prior analysis
Consensus	~£0.03	instant	Cross-user aggregate landscape (k>=3 checks)
Quick	~£0.07	~15s	Web search + heuristic classification
Full	~£0.15	~60-90s	web + specialist APIs, LLM classification, element decomposition

Three payment rails: x402 (USDC/SIWE), Skyfire (JWT), prepaid credits.

MCP Server

json

{
  "mcpServers": {
    "tru8": {
      "command": "python",
      "args": ["-m", "tru8_mcp"],
      "env": { "TRU8_API_KEY": "tru8_sk_..." }
    }
  }
}

Tools: tru8_check (evidence research with tier fallback), tru8_get_result (with analytics), tru8_get_result_raw.

Manifest Verification

Every completed check receives an HMAC-SHA256 signed manifest. Verify the signed fields haven't changed since signing:

bash

curl https://api.trueight.com/verify/{check_id}

Testing

bash

cd backend
pytest tests/ -v                        # Full suite (990 tests)
pytest tests/unit/pipeline/ -v          # Pipeline unit tests
pytest tests/unit/agent/ -v             # Agent/commerce tests
pytest tests/integration/ -v            # Integration tests

Infrastructure

Service	Port	Purpose
PostgreSQL 16	5433	Primary database (SQLModel)
Redis 7	6379	Cache + rate limiting
Qdrant	6333	Vector similarity search
MinIO (optional)	9000	Local S3 for file uploads

Auth: Clerk (JWT + JWKS) for dashboard, API keys for developer access. Payments: Stripe (4 subscription tiers) + Agent per-use payments.

API Documentation

Swagger UI: http://localhost:8000/api/docs
ReDoc: http://localhost:8000/api/redoc
Health: http://localhost:8000/api/v1/health/

Environment Variables

Copy backend/.env.example to backend/.env. Key groups:

Group	Required	Variables
Database	Yes	`DATABASE_URL`
Redis	Yes	`REDIS_URL`
Auth	Yes	`CLERK_SECRET_KEY`, `CLERK_JWT_ISSUER`
LLM	Yes	`GOOGLE_AI_API_KEY`, `OPENAI_API_KEY`
Search	Yes (at least one)	`SERPER_API_KEY`, `BRAVE_API_KEY`, `SERP_API_KEY`
Payments	For subscriptions	`STRIPE_SECRET_KEY`, `STRIPE_WEBHOOK_SECRET`
Government APIs	Optional	`FRED_API_KEY`, `NOAA_API_KEY`, etc.
Manifest signing	Optional	`MANIFEST_SIGNING_ENABLED`, `MANIFEST_SIGNING_KEY`

Feature flags (all default to sensible values): ENABLE_API_RETRIEVAL, ENABLE_LLM_RELEVANCE_SCORER, ENABLE_QUERY_PLANNING, etc. See backend/.env.example for the full list.

claim	string	required	A factual claim ("The Earth's average temperature rose 1.1°C since 1880") or an article URL (https://example.com/article). URLs are auto-detected and the pipeline extracts claims from the page content.
max_tier	string	optional	Maximum tier to attempt — "lookup", "consensus", "quick" or "full" (default). A CEILING, not a floor: cached and consensus hits still return instantly at their own lower price, so the default costs ~£0.15 only for claims never researched before. Set "quick" to cap spend at ~£0.07, accepting web-search-only sourcing and heuristic classification (see _meta.limitations).
max_age_hours	any	optional	Skip cache hits older than this many hours. Stale lookup hits are discarded and the pipeline re-runs at the next tier up to max_tier. 0 means never serve a cached result.
compact	boolean	optional	If true, strip the full evidence arrays and computed analytics from the response, leaving claims and claim maps. Smaller payload for agents that only need orientation.

Tru8 Evidence Research

io.github.SamYatesSmith/tru8 — Model Context Protocol (MCP) Server

🛠️ Key Features

🚀 Use Cases

⚡ Developer Benefits

⚠️ Limitations

Topics

tru8_check

tru8_get_result

tru8_get_result_raw

Tru8 — AI-Powered Evidence Research Platform

Quick Start

Prerequisites

1. Start Infrastructure

2. Backend

3. Frontend

Project Structure

Pipeline

Six Evidence Views

API & Agent Access

Agent Commerce Gateway

MCP Server

Manifest Verification

Testing

Infrastructure

API Documentation

Environment Variables

Licence

Install

Configuration

Related servers