brunoamaral/gregory-ai/gregory-api-search

GregoryAI API Search

GregoryAI aggregates clinical research — ~49,000 articles and ~29,000 clinical trials — tagged by team, research subject, and category, with machine-learning relevance scores. This skill queries it read-only over plain HTTP. No API key or login is required for reads.

Base URL: https://api.brain-regeneration.com

Golden rules

• Read-only. Only ever issue GET requests. Never suggest or attempt to create, edit, or delete anything.
• Ask for JSON. Append ?format=json (or send header Accept: application/json) or you may get an HTML page.
• Results are paginated. Default 10 per page, max page_size=100. Response wraps results in

count, next, previous, current_page, total_pages, page_size, results. Follow the next URL to page, or add all_results=true to return everything (use sparingly — some result sets are huge).

• CSV export: add format=csv&all_results=true instead of format=json.
• Some collections are large (48k+ articles, 248k+ authors) — always narrow with filters before fetching.

How to call it

Use curl (preferred for scripting/paging) or WebFetch. Every example below is a complete URL.

curl -s "https://api.brain-regeneration.com/articles/?search=microglia&page_size=5&format=json"

The one search rule you must know

The search= parameter (on /articles/ and /trials/) runs a boolean search over title + summary (article abstracts for /articles/, trial summaries for /trials/):

• Space-separated terms are AND-ed: search=stem cells regeneration
• OR (uppercase) for alternatives: search=alzheimer OR parkinson
• -term or NOT term to exclude: search=covid -vaccine
• "quoted phrases" match contiguously: search="multiple sclerosis"
• (parentheses) group: search=(microglia OR astrocyte) AND inflammation

URL-encode it: spaces → %20 (or +), quotes → %22. See references/search-syntax.md.

Endpoint map

Discover the topics first (subjects & teams)

Most filtering keys off subject (a research topic) and team. Look them up first — every subject row carries its team_id:

curl -s "https://api.brain-regeneration.com/subjects/?format=json"

Subjects on this instance (subject_id → topic). Pass subject_id to /articles/ and /trials/ to scope a search to that research area:

(IDs can change and new subjects may be added — re-fetch /subjects/ to confirm the current list.)

Quick-start recipes

BASE="https://api.brain-regeneration.com"
 
# 1. Recent articles mentioning microglia
curl -s "$BASE/articles/?search=microglia&ordering=-published_date&page_size=10&format=json"
 
# 2. AI-relevant Multiple Sclerosis articles (subject 1), most confident first
curl -s "$BASE/articles/?subject_id=1&relevant=true&ordering=-ml_score&format=json"
 
# 3. Article by DOI
curl -s "$BASE/articles/?doi=10.1007/s00296-026-06233-x&format=json"
 
# 4. Recruiting Parkinson's trials (subject 14)
curl -s "$BASE/trials/?subject_id=14&status=Recruiting&format=json"
 
# 5. A trial by its NCT number
curl -s "$BASE/trials/?nct=NCT04305002&format=json"
 
# 6. Find an author by name, see their article count
curl -s "$BASE/authors/?full_name=smith&sort_by=article_count&order=desc&format=json"
 
# 7. Alzheimer's articles published in 2024, as CSV
curl -s "$BASE/articles/?subject_id=13&published_date_after=2024-01-01&published_date_before=2024-12-31&format=csv&all_results=true"

Reference files

Read the relevant reference file before building anything non-trivial — each lists every supported filter, the exact response fields, and worked examples:

• references/articles.md — all article filters, ML relevance, dates, ordering
• references/trials.md — registry-ID lookups, conditions, sponsors, results
• references/authors.md — author search and their publication lists
• references/subjects-and-categories.md — topic & category discovery
• references/search-syntax.md — full boolean search= grammar
• references/rss-feeds.md — author and trial RSS feeds