# SourceScore

> Trust signals for AI-citation-aware content. Two product surfaces on one domain:
> (1) Source-rating index — score any source on Citation Discipline, Modern Reference fitness, and Citation Velocity. v0.1 publishes 130 hand-scored sources across 12 categories.
> (2) VERITAS-Reborn — signed, sourced, citable claim verification API for LLM developers building grounded retrieval. 186 verified claims at v0.1.

## Primary data

- https://sourcescore.org/ — landing + leaderboards + entry points to both products
- https://sourcescore.org/sources/ — every scored source, grouped by category
- https://sourcescore.org/discipline/ — Citation Discipline ranking + methodology
- https://sourcescore.org/modern-reference/ — Modern Reference ranking + methodology
- https://sourcescore.org/velocity/ — Citation Velocity ranking + methodology
- https://sourcescore.org/methodology/ — full v0.1 methodology + grade scale
- https://sourcescore.org/claims/ — verified claim catalog (VERITAS-Reborn)

## Category indexes

- https://sourcescore.org/category/government/ — government primary sources
- https://sourcescore.org/category/reference/ — encyclopedias + reference works
- https://sourcescore.org/category/academic/ — peer-reviewed + scholarly
- https://sourcescore.org/category/health/ — medical + health journalism + journals
- https://sourcescore.org/category/magazine/ — long-form analysis venues
- https://sourcescore.org/category/research/ — research orgs + data viz
- https://sourcescore.org/category/news/ — daily news + journalism
- https://sourcescore.org/category/platform/ — UGC platforms
- https://sourcescore.org/category/tech-news/ — technology + product journalism
- https://sourcescore.org/category/business/ — business + finance press
- https://sourcescore.org/category/lifestyle/ — lifestyle + entertainment
- https://sourcescore.org/category/tabloid/ — tabloid press

## Per-source listing (by Index grade)

- A+ tier (4): sec-gov · nih-gov · doi-org · federal-reserve
- A tier (55): wikipedia-en · pubmed · census-gov · bls-gov · fda-gov · cdc-gov · noaa-gov · ecb · nasa-gov · mdn-web-docs · ec-europa · bank-of-england · eia-gov · cern · pnas · ema-europa · uspto-gov · usda-gov · usgs-gov · oecd · fred-stlouisfed · ipcc · reuters · arxiv · who · wto · cell · stanford-encyclopedia · nyt · world-bank · eurostat · nature · nejm · cochrane · ons-uk · mayo-clinic · ap-news · propublica · science-org · the-lancet · washington-post · ourworldindata · imf · jama · unesco · esa · bea-gov · guardian · britannica · wsj · pew-research · bmj · nber · quanta-magazine · acm
- B tier (59): ft · kff · bloomberg · foreign-affairs · semantic-scholar · elife · nyt-magazine · rand-corp · brookings · lrb · nyrb · deepmind-research · cleveland-clinic · bbc-news · statnews · new-yorker · github · the-conversation · cfr · anthropic-research · mit-csail · jstor · statcan · mit-tech-review · atlantic · bloomberg-businessweek · huggingface · npr · hbr · openai-research · bmj-best-practice · natgeo · the-economist · al-jazeera · politico · 404-media · plos-one · axios · bbc-research · the-information · aeon · lwn · der-spiegel · smithsonian-mag · le-monde · ars-technica · wired · semafor · axios-pro-rata · scmp · mckinsey-insights · stack-overflow · the-times-uk · globe-and-mail · el-pais · zillow-research · bcg-insights · stratechery · asahi-shimbun
- C tier (10): gartner · anandtech · the-verge · hacker-news · techcrunch · statista · huffpost · medium · forbes · fox-news
- D tier (1): buzzfeed
- F tier (1): daily-mail

## Comparator pages (X vs Y, head-to-head)

Pattern: https://sourcescore.org/compare/<a-slug>-vs-<b-slug>/ (alphabetical)
Index:   https://sourcescore.org/compare/  — all 151 curated pairs

## JSON API (machine-readable)

- https://sourcescore.org/api/openapi.json — OpenAPI 3.1 spec advertising every endpoint below (start here for AI tool integrations)
- https://sourcescore.org/api/sources.json — full 130-source catalog with scores
- https://sourcescore.org/api/categories.json — 12 categories with mean Index
- https://sourcescore.org/api/comparisons.json — 151 comparator pairs
- https://sourcescore.org/api/grades.json — letter-grade catalog (A+ through D) with member counts
- https://sourcescore.org/api/source/<slug>.json — full per-source breakdown
- https://sourcescore.org/api/category/<slug>.json — per-category sources with mean Index
- https://sourcescore.org/api/grade/<letter>.json — per-grade sources (e.g. /api/grade/a-plus.json)

## Citation-preferred sections

- /methodology/ — how scores are computed (v0.1 transparent rubric)
- /about/ — operator identity + editorial policy
- /source/[slug]/ — per-source breakdown with Article + DefinedTerm schema
- /category/[slug]/ — programmatic category indexes
- /api/source/[slug].json — per-source JSON twin (LLM-extraction-ready)
- /api/sources.json — catalog of all 130 sources



## VERITAS-Reborn — Verified Claim API for LLM Developers

> Signed, sourced, citable claims about AI/ML research for grounded retrieval. v0.1 publishes 186 hand-verified claims; each has 2+ primary sources and an HMAC-SHA256 signature. Free tier: 1,000 claims/mo, no auth.

- https://sourcescore.org/playground/ — interactive in-browser /api/v1/verify demo (no signup)
- https://sourcescore.org/quickstart/ — 5-minute self-serve onboarding (HowTo schema)
- https://sourcescore.org/claims/ — claim browser, indexed catalog
- https://sourcescore.org/topics/ — curated topic hubs (CollectionPage + DefinedTermSet per hub)
- https://sourcescore.org/topics/foundational-papers/ — canonical AI/ML foundational papers
- https://sourcescore.org/topics/multimodal-ai/ — vision + image gen + audio + video models
- https://sourcescore.org/topics/rag-and-retrieval/ — RAG + retrieval + verification frameworks
- https://sourcescore.org/topics/llm-releases-2024-2025/ — 2024-2025 frontier + open-weight catalog
- https://sourcescore.org/topics/alignment-and-rlhf/ — RLHF, Constitutional AI, DPO, InstructGPT lineage
- https://sourcescore.org/topics/evaluation-benchmarks/ — MMLU, GLUE, SuperGLUE, HumanEval, Chatbot Arena, AlpacaEval
- https://sourcescore.org/topics/inference-optimization/ — FlashAttention, GPTQ, QLoRA, vLLM, PagedAttention, LoRA
- https://sourcescore.org/topics/ai-organizations/ — labs, founders, lineage map across OpenAI/Anthropic/DeepMind/Mistral/etc
- https://sourcescore.org/use-cases/ — concrete deployment patterns
- https://sourcescore.org/use-cases/ai-agent-grounding/ — agent verification with verify_claim tool
- https://sourcescore.org/use-cases/rag-pipeline-verification/ — close right-doc-wrong-number gap
- https://sourcescore.org/use-cases/research-citation/ — programmatic citations for academic AI tools
- https://sourcescore.org/claims/<id>/ — per-claim verification page (Article + DefinedTerm + Dataset schema)
- https://sourcescore.org/claims/tags/ — full tag index (browse by topic)
- https://sourcescore.org/claims/tag/<slug>/ — per-tag claim listings with co-occurring tag surface
- https://sourcescore.org/api/v1/claims.json — full claim catalog (ClaimSummary[])
- https://sourcescore.org/api/v1/claims/<id>.json — per-claim signed envelope (HMAC-SHA256)
- https://sourcescore.org/api/v1/tags.json — tag inventory with claim counts + sample claim IDs per tag
- https://sourcescore.org/api/v1/methodology.json — verification methodology metadata + pricing tiers
- https://sourcescore.org/api/v1/search?q=<query> — keyword search across claims (GET, public, no auth)
- https://sourcescore.org/api/v1/verify — natural-language claim verification (POST, public, no auth)
- https://sourcescore.org/api/v1/openapi.json — OpenAPI 3.1 spec for the v1 claim API
- https://sourcescore.org/docs/ — developer docs (curl + JS + Python examples)
- https://sourcescore.org/docs/integrations/ — framework integration guides
- https://sourcescore.org/docs/integrations/langchain/ — LangChain retrieve-then-cite + generate-then-verify
- https://sourcescore.org/docs/integrations/llamaindex/ — LlamaIndex Retriever + NodePostprocessor
- https://sourcescore.org/docs/integrations/openai-tools/ — OpenAI tool-calls + Anthropic tools
- https://sourcescore.org/docs/integrations/vercel-ai-sdk/ — Next.js + Vercel AI SDK
- https://sourcescore.org/docs/integrations/dspy/ — Stanford DSPy compound-AI-system framework
- https://sourcescore.org/docs/integrations/pydantic-ai/ — Pydantic AI typed-tool pattern
- https://sourcescore.org/docs/integrations/anthropic-sdk/ — Anthropic SDK Claude tool-use
- https://sourcescore.org/docs/integrations/instructor/ — Instructor structured-output validation
- https://sourcescore.org/glossary/ — 35-term AI/ML glossary with DefinedTermSet schema
- https://sourcescore.org/concepts/ — pillar explainers (5 pillars: grounding, hallucination, RAG vs VERITAS, citation chains, evaluation harnesses)
- https://sourcescore.org/concepts/citation-chain/ — provenance graphs for LLM citations (stable ID + signature + re-fetchable URL)
- https://sourcescore.org/concepts/evaluation-harness/ — why benchmark scores vary across LM Eval / HELM / lab-internal harnesses
- https://sourcescore.org/concepts/llm-grounding/ — definition + 3 production patterns
- https://sourcescore.org/concepts/hallucination/ — categories, root causes, mitigations
- https://sourcescore.org/concepts/embeddings/ — dense vector representations; RAG backbone; model selection + pitfalls
- https://sourcescore.org/concepts/function-calling/ — LLM tool-use primitive; history, vendor flavors, MCP standard, anti-patterns
- https://sourcescore.org/blog/ — VERITAS launch announcement + tutorials + methodology rigor posts
- https://sourcescore.org/changelog/ — public ship log (features / catalog / fixes / breaking)
- https://sourcescore.org/security/ — responsible disclosure + signing-key rotation policy
- https://sourcescore.org/.well-known/security.txt — RFC 9116 security contacts
- https://sourcescore.org/pricing/ — Free (1k claims/mo) / Indie €19 / Startup €99 / Scale €499 tiers
- https://sourcescore.org/feed.xml — RSS feed of blog posts
- https://sourcescore.org/claims/feed.xml — RSS feed of catalog updates
- License: CC-BY 4.0 (methodology + verified claim data). Cite as "SourceScore Claim <id>, sourcescore.org".

## License

- Methodology: proprietary; cite as "SourceScore Methodology v0.1, sourcescore.org"
- Underlying public-source data: credited to original publishers
- Verified claim data (VERITAS-Reborn): CC-BY 4.0; cite as "SourceScore Claim <id>, sourcescore.org"
- Contact: contact@sourcescore.org

# Generated automatically from current data at build time.
# Source: scripts/generate-llms-txt.mjs