AstroBridge

Version: 0.3.3 — Live Catalog Lookup, Multi-Catalog AI Identification, 261 Tests

AstroBridge is an astronomical source matching and cross-catalog identification pipeline. You give it an object name or sky coordinates; it fans out to real catalogs (SIMBAD, NED, Gaia DR3, 2MASS), merges the results, and optionally generates a plain-language description using an LLM.

---

Install

git clone https://github.com/myrakhandelwal/AstroBridge.git
cd AstroBridge
python -m venv .venv && source .venv/bin/activate

pip install -e .[dev]        # core + tests
pip install -e .[live]       # + Gaia DR3, 2MASS, SIMBAD/NED live TAP
pip install -e .[web]        # + FastAPI web console
pip install anthropic         # + Claude AI descriptions (optional)
pip install openai            # + OpenAI/local-compatible AI descriptions (optional)

---

Quick Start

# End-to-end pipeline demo (synthetic data, no network needed)
python demo.py

# Identify an object and get catalog recommendations
astrobridge-identify "Proxima Centauri"
astrobridge-identify "Find nearby red dwarf stars"

# Interactive menu-driven demo
python interactive_demo.py

# Web console at http://127.0.0.1:8000
astrobridge-web

---

Live Catalog Lookup

lookup_object() uses a two-step strategy — no local database required:

1. Name resolution — fans out to SIMBAD + NED concurrently
2. Positional enrichment — uses the returned coordinates to cone-search Gaia DR3 and 2MASS

import asyncio
from astrobridge.lookup import lookup_object, lookup_by_coordinates

async def main():
    # Look up by name
    obj = await lookup_object("Proxima Centauri")
    if obj:
        print(obj.primary_name)          # Proxima Centauri
        print(f"RA={obj.ra:.4f}  Dec={obj.dec:.4f}")
        print("catalogs:", list(obj.catalog_entries.keys()))
        print("photometry:", obj.photometry_summary)

    # Cone search by coordinates
    results = await lookup_by_coordinates(ra=217.429, dec=-62.680, radius_arcsec=60)
    for r in results:
        print(r.primary_name, r.ra, r.dec)

asyncio.run(main())

Requires pip install -e .[live] for live network queries. Falls back to local connectors automatically when pyvo is not installed.

---

Object Identification

Combines NLP routing, live catalog lookup, and optional AI description in one call:

import asyncio
from astrobridge.identify import identify_from_catalogs

async def main():
    result = await identify_from_catalogs("M31")
    print(result["object_class"])       # galaxy
    print(result["description"])        # plain-language description
    print(result["top_catalogs"])       # ['NED', 'SDSS', 'ALLWISE']
    print(result["catalog_data"])       # real RA/Dec + photometry from catalogs

asyncio.run(main())

For CLI use:

astrobridge-identify "M31"
astrobridge-identify "Find high-redshift quasars"
astrobridge-identify --json "Betelgeuse"

---

AI Descriptions

Set environment variables to enable real LLM descriptions. Defaults to a deterministic stub (no API key needed).

# Anthropic Claude (recommended)
export AI_PROVIDER=anthropic
export AI_API_KEY=sk-ant-...

# OpenAI
export AI_PROVIDER=openai
export AI_API_KEY=sk-...
export AI_MODEL=gpt-4o-mini

# Local OpenAI-compatible endpoint (example)
export AI_PROVIDER=local
export AI_BASE_URL=http://127.0.0.1:11434/v1
export AI_MODEL=your-local-model

# Stub (default — no key needed)
export AI_PROVIDER=stub

Configuration is considered "real AI" when the provider requirements are satisfied:

• openai or anthropic: set AI_API_KEY
• local: set AI_BASE_URL
• stub: deterministic offline fallback

from astrobridge.ai_description import generate_description
from astrobridge.models import UnifiedObject

obj = UnifiedObject(primary_name="M31", ra=10.685, dec=41.269, object_type="galaxy")
print(generate_description(obj))

---

Bayesian Cross-Matching

from astrobridge.matching import BayesianMatcher

matcher = BayesianMatcher(proper_motion_aware=True)
matches = matcher.match(reference_sources, candidate_sources)

for m in matches:
    print(f"{m.source1_id} ↔ {m.source2_id}  p={m.match_probability:.3f}  sep={m.separation_arcsec:.2f}\"")

Matching controls available on every QueryRequest:

Parameter	Type	Description
proper_motion_aware	bool	Apply proper-motion corrections across epochs
match_epoch	datetime	Reference epoch for PM corrections
weighting_profile	str	balanced | position_first | photometry_first
astrometric_weight	0–1	Manual astrometric weight override
photometric_weight	0–1	Manual photometric weight override

---

Orchestrator (Multi-Catalog Queries)

import asyncio
from astrobridge.api import AstroBridgeOrchestrator, QueryRequest
from astrobridge.matching import BayesianMatcher
from astrobridge.routing import NLPQueryRouter

async def main():
    orch = AstroBridgeOrchestrator(router=NLPQueryRouter(), matcher=BayesianMatcher())

    # Name query
    req = QueryRequest(query_type="name", name="Sirius", auto_route=True)
    resp = await orch.execute_query(req)
    print(resp.status, len(resp.sources), "sources")

    # Natural language
    req2 = QueryRequest(
        query_type="natural_language",
        description="Find nearby infrared-bright red dwarfs",
        auto_route=True,
        weighting_profile="position_first",
    )
    resp2 = await orch.execute_query(req2)
    print(resp2.routing_reasoning)

asyncio.run(main())

---

Supported Catalogs

Catalog	Live Adapter	Name Lookup	Cone Search	Best For
SIMBAD	SimbadTapAdapter	✅	✅	All objects, name resolution
NED	NedTapAdapter	✅	✅	Galaxies, AGN, quasars
Gaia DR3	GaiaDR3TapAdapter	—	✅	Stars: astrometry + proper motions
2MASS	TwoMassTapAdapter	—	✅	Stars + galaxies: J/H/Ks photometry
SDSS	routing only			Galaxies, QSOs, optical photometry
WISE	routing only			Mid-IR sources, AGN
AllWISE	routing only			Improved WISE + proper motions
PanSTARRS	routing only			Transients, wide-field optical
ZTF	routing only			Supernovae, variables, transients
ATLAS	routing only			Transient alerts, SNe
Hipparcos	routing only			Bright stars (V < 12)
VizieR	routing only			Any published catalog table
NASA Exoplanet Archive	routing only			Exoplanet host stars

Live adapters require pip install -e .[live].

---

Web Console

pip install -e .[web]
astrobridge-web
# → http://127.0.0.1:8000

REST API endpoints:

Method	Endpoint	Description
POST	/api/query	Run a cross-catalog query
POST	/api/identify	Identify object + live catalog lookup + AI description
POST	/api/jobs	Submit async background job
GET	/api/jobs/{id}	Check job status
GET	/api/jobs/{id}/result	Fetch completed job result
POST	/api/analytics/event	Record analytics event
GET	/api/analytics/summary	Aggregate analytics summary
POST	/api/benchmark/run	Run latency benchmark

---

Quality Gates

ruff check .          # linting
mypy astrobridge/     # strict type checking on core modules
pytest -q             # 261 tests, zero warnings

All gates run automatically on every push via GitHub Actions.

---

Persistent State

Jobs and analytics are persisted in SQLite. Default path: .astrobridge/state.db

export ASTROBRIDGE_STATE_DB="/path/to/state.db"   # override location

---

Package API Reference

Import	Description
astrobridge.lookup.lookup_object	Two-step live catalog fan-out by name
astrobridge.lookup.lookup_by_coordinates	Concurrent cone search across all live adapters
astrobridge.identify.identify_from_catalogs	NLP routing + live lookup + AI description
astrobridge.identify.identify_object	NLP classification only (no network)
astrobridge.api.AstroBridgeOrchestrator	Multi-catalog query orchestration
astrobridge.api.QueryRequest	Request model with matcher + routing controls
astrobridge.matching.BayesianMatcher	Probabilistic cross-matching with PM support
astrobridge.matching.ConfidenceScorer	Match confidence with weighting profiles
astrobridge.connectors.SimbadTapAdapter	Live SIMBAD TAP (name + cone)
astrobridge.connectors.NedTapAdapter	Live NED TAP (name + cone)
astrobridge.connectors.GaiaDR3TapAdapter	Live Gaia DR3 TAP (cone only)
astrobridge.connectors.TwoMassTapAdapter	Live 2MASS TAP via IRSA (cone only)
astrobridge.ai_description.generate_description	LLM description (anthropic / openai / stub)
astrobridge.analytics.AnalyticsStore	SQLite-backed telemetry events
astrobridge.jobs.JobManager	Background job lifecycle manager
astrobridge.benchmarking.BenchmarkRunner	Reproducible latency benchmarks
astrobridge.database	SQLite persistence for objects, sources, calibration frames
astrobridge.ccd_calibration.calibrate_ccd	CCD reduction: bias/dark/flat (astropy or NumPy)
astrobridge.models.UnifiedObject	Merged multi-catalog view with from_sources()

---

Documentation

Document	Description
docs/Command Guide.md	CLI commands, REST API, Python API examples
docs/Algorithm and Science.md	Bayesian matching math, proper-motion corrections
docs/Architecture Guide.md	System design, component flow, custom adapters
docs/Deployment Guide.md	Docker, AWS, PyPI release, monitoring
docs/Platforms_and_Catalogs_Matrix.md	All 13 catalogs: routing scores, adapter status
docs/Test Suite Guide.md	All 261 tests described by file and category
docs/RELEASE_NOTES.md	Full changelog: v0.1.1 → v0.3.3

---

Project Layout

astrobridge/
├── models.py           # Source, Coordinate, UnifiedObject, MatchResult
├── lookup.py           # Live two-step catalog fan-out
├── identify.py         # NLP classification + identify_from_catalogs()
├── connectors.py       # SimbadTapAdapter, NedTapAdapter, GaiaDR3TapAdapter, TwoMassTapAdapter
├── ai_description.py   # LLM descriptions (anthropic / openai / stub)
├── database.py         # SQLite persistence layer
├── ccd_calibration.py  # CCD reduction pipeline
├── geometry.py         # Angular distance calculations
├── routing/            # NLPQueryRouter, CatalogRanker, 13 CatalogTypes
├── matching/           # BayesianMatcher, ConfidenceScorer, SpatialIndex
├── api/                # AstroBridgeOrchestrator, QueryRequest/Response
├── web/                # FastAPI app, REST endpoints, HTML console
├── analytics.py        # AnalyticsStore, event tracking
├── jobs.py             # JobManager, background query lifecycle
└── benchmarking.py     # BenchmarkRunner

---

Release History

Tag	Date	Highlights
v0.3.3	Apr 14 2026	Live catalogs, Gaia DR3 + 2MASS adapters, AI descriptions, 261 tests
v0.3.2	Apr 9 2026	mypy strict fixes
v0.3.1	Apr 9 2026	Interactive demo, test suite guide
v0.3.0	Apr 9 2026	PEP 621 packaging, ruff, async concurrency
v0.2.0	Apr 8 2026	Comprehensive docs, deployment guide
v0.1.1	—	Initial release

---

License

MIT License — Copyright © 2026 Myra Khandelwal

Simbad and NED fall back to local deterministic datasets when pyvo is not installed, so all tests and demos run without network access.