ASTER -- Agentic Science Toolkit for Exoplanet Research
Authors: Emilie Panek, Alexander Roman, Gaurav Shukla, Leonardo Pagliaro, Katia Matcheva, Konstantin Matchev
Organizations: Department of Physics and Astronomy, University of Alabama, Tuscaloosa, AL, 35487, USA · Department of Physics and Astronomy ‘Galileo Galilei’, University of Padova, Vicolo dell’Osservatorio 3, I-35122 Padova, Italy
The expansion of exoplanet observations has created a need for flexible, accessible, and user-friendly workflows. Transmission spectroscopy has become a key technique for probing atmospheric composition of transiting exoplanets. The analyses of these data require the combination of archival queries, literature search, the use of radiative transfer models, and Bayesian retrieval frameworks, each demanding specialized expertise. Modern large language models enable the coordinated execution of complex, multi-step tasks by AI agents with tool integration, structured prompts, and iterative reasoning. In this study we present ASTER, an Agentic Science Toolkit for Exoplanet Research. ASTER is an orchestration framework that brings LLM capability to the exoplanetary community by enabling LLM-driven interaction with integrated domain-specific tools, workflow planning and management, and support for common data analysis tasks. Currently ASTER incorporates tools for downloading planetary parameters and observational datasets from the NASA Exoplanet Archive, as well as the generation of transit spectra from the TauREx radiative transfer model, and the completion of Bayesian retrieval of planetary parameters with TauREx. Beyond tool integration, the agent assists users by proposing alternative modeling approaches, reporting potential issues and suggesting solutions, and interpretations. We demonstrate ASTER's workflow through a complete case study of WASP-39b, performing multiple retrievals using observational data available on the archive. The agent efficiently transitions between datasets, generates appropriate forward model spectra and performs retrievals. ASTER provides a unified platform for the characterization of exoplanet atmospheres. Ongoing development and community contributions will continue expanding ASTER's capabilities toward broader applications in exoplanet research.
Figures & tables
Figure 1: Schema of internal workings of the agent object. The agent comprises the LLM, the tools and the context. Figure adapted from Roman and Roman (2026) .
Figure 2: Example of the Orchestral interface. The LLM provider can be changed in the bottom left corner, below the space where users write their messages. The cost tracking can be seen in the top right corner.
Molecule
Reference
H 2 O
Polyansky et al. (2018) ; Chubb et al. (2021)
CO
Li et al. (2015) ; Chubb et al. (2021)
CO 2
Yurchenko et al. (2025)
CH 4
Yurchenko et al. (2024)
NH 3
Coles et al. (2019) ; Chubb et al. (2021)
HCN
Barber et al. (2013) ; Chubb et al. (2021)
Table 1: Sources of cross-section data used in this work, downloaded from the Exomol website.
Figure 5: A plot generated, formatted, and displayed by the agent during the example conversation. It represents the transmission spectrum for WASP-39b, calculated by the agent and shown using the DisplayImage tool.
Figure 6: A plot generated, formatted, and displayed by the agent during the example conversation. It represents the best-fit model (orange curve) retrieved by the agent on WASP-39b NIRSpec observation (blue points).
Parameter
Value
Stellar mass
0.913 M ⊙
Stellar radius
0.939 R ⊙
Stellar effective temperature
5485 K
Planet mass
0.281 M jup
Planet radius
1.279 R jup
Planet equilibrium temperature
1166 K
Table 2: Planetary and stellar parameters for WASP-39b used in this study. Values were obtained automatically by the agent and verified against the NASA Exoplanet Archive.
Figure 7: Transmission spectrum at native resolution in the background (gray points), overlaid with a binned spectrum (red curve) recalculated by the agent. Prompt: Can you bin down the spectrum and plot both resolutions on a same plot?
Figure 8: Transmission spectrum at native resolution in the background (gray points), overlaid with a binned spectrum recalculated by the agent. Prompt: Can you now bin the forward model down to JWST medium resolution?
Figure 9: Transmission spectrum at native resolution in the background (gray points), with a binned down spectrum recalculated by the agent on top. Prompt: Can you now bin the forward model down to Ariel resolution?
Figure 10: Comparison of all nine observations for WASP-39b with NIRSpec. These observations are public and are published in JWST Transiting Exoplanet Community Early Release Science Team et al. (2023) ; Alderson et al. (2023) ; Rustamkulov et al. (2023) ; Carter et al. (2024) . This figure was produced by the agent. Prompt: Can you plot all the observations in a same figure for comparison?
Fitted Parameter
Retrieval 1
Retrieval 2
T
1318.5
666.5
Rp
1.29
1.30
logH2O
-8.5
-4.0
logCH4
-8.8
-7.1
logCO
-8.0
-5.7
logCO2
-8.7
-5.0
Table 3: Fitted parameters from the two retrieval runs. The results are discussed in Section 5 .
Figure 11: Comparison of the two retrieval results launched by the agent. The first observation used by the agent ( JWST Transiting Exoplanet Community Early Release Science Team et al., 2023 ) and the corresponding retrieved spectrum are shown in blue. The second observation used by the agent ( Rustamkulov et al., 2023 ) and the corresponding retrieved spectrum are shown in red. Prompt: Can you do a plot with the two observations and the two retrieved spectra so I can compare the results?
Figure 12: Corner plot of the posterior distributions of the fitted parameters from the retrievals launched by the agent. The blue distributions are corresponding to the first retrieval on JWST Transiting Exoplanet Community Early Release Science Team et al. (2023) data. The red distributions are corresponding to the second retrieval on Rustamkulov et al. (2023) data. Prompt: Can you now do a corner plot with the probability distribution results of the two retrievals on the same corner plot figure?
Real-world Earth observation (EO) agents must translate high-level scientific questions into executable workflows to acquire observations, prepare data, perform domain computations, and derive conclusions from runtime evidence. Existing EO agents typically start from supplied observations, while benchmarks typically provide prepared inputs or candidate answers, leaving full-chain open-world EO execution largely untested. We present Earth-Agent-Pro, an execution-adaptive Plan-and-Execute framework using expert-authored skills to constrain planning and runtime tool use. Workflow-centered structured memory records planned steps, accepted evidence, and their dependencies, enabling repair of only the affected workflow suffix when runtime evidence invalidates a step. Separate large language model adapters use sequence-level supervised fine-tuning for planner workflow composition and node-level group relative policy optimization with locally verifiable rewards for executor tool-argument grounding. Earth-Bench-Pro instantiates 248 expert-curated task cores as 744 questions under three matched regimes. Its 248 Open-World Execution questions span RGB imagery, spectral observations, and remote sensing products, pairing high-level requests with runtime data requirements, executable trajectories, and open-ended answers grounded in execution evidence. With a shared GPT-5 backbone, Earth-Agent-Pro achieves 66.13% LLM-as-Judge accuracy, exceeding ReAct by 20.95 points in this metric and 24.44 points in Tools-In-Order. Joint adapter tuning raises Qwen3.5-9B LLM-as-Judge accuracy from 38.31% to 50.00%, an 11.69-point gain over the untuned configuration. Planning-only evaluation and execution with the reference workflow show that the adapters improve workflow composition and argument grounding, respectively. Code and datasets will be released soon.
Zhutao Lv, Chenhao Dang, Yi Feng +5
Tsinghua Shenzhen International Graduate School, Tsinghua University · Sun Yat-Sen University · Shanghai Jiao Tong University +3
I present EXOVEIL, a transit detection system that learns what a star's brightness should look like and flags when reality disagrees. Unlike existing systems that require phase-folded input, EXOVEIL operates on raw flux time series and can detect planets that transit only once.A Transformer world model, trained on 16,499 Kepler light curves with transit-masked self-supervised learning, predicts expected stellar flux. A matched-filter detector with variance weighting extracts transit signals from the prediction residuals. A learned classifier (XGBoost) separates planets from false positives, achieving AUC 0.938 on Kepler DR25. Applied to single-transit injection-recovery, EXOVEIL recovers 32% of transits at 1000 ppm depth a task where all classification-based systems score 0% by construction. A blind search of 3,737 Kepler stars yields 179 new transit-like signals not present in the DR25 TCE catalogue, including 46 monotransit candidates. Applied withoutretraining to 47 confirmed TESS planets in the PLATO LOPS2 field, EXOVEIL achieves 100% recovery, demonstrating zero-shot cross-mission transfer. At PLATO's 25-second cadence, detection reaches 100 ppm -- approaching the Earth-analog regime. I provide the first application of conformal prediction to transit detection (95.9% empirical coverage) and release the system as pip install exoveil with pretrained weights and a candidate catalogue.
The detection and atmospheric characterization of exoplanets have entered a new data-intensive era driven by the James Webb Space Telescope and the upcoming Ariel mission. Modern surveys produce millions of light curves and high-resolution spectra that overwhelm traditional pipelines, motivating the rapid integration of Machine Learning and Deep Learning methods into the exoplanet workflow. This review synthesizes the latest progress in applying ML/DL techniques to exoplanet detection (transit identification, candidate vetting, false-positive rejection) and atmospheric characterization (retrieval, detrending, cross-correlation, surrogate modelling) in the context of JWST and Ariel. We start with classical algorithms such as Random Forests and Convolutional Neural Networks, move through Transformers and Recurrent architectures, then survey modern simulation-based inference using Neural Posterior Estimation and Flow Matching Posterior Estimation with normalizing or continuous normalizing flows. We discuss benchmark efforts, including the Ariel Machine Learning Data Challenges (2019 to 2025) hosted with NeurIPS, and key JWST case studies such as the WASP-39b Early Release Science programme. Results indicate that DL approaches consistently match or exceed traditional pipelines in both speed and accuracy, while ML-driven retrievals reduce inference time from CPU-hours to seconds and can accelerate nested-sampling retrievals by factors of 3-8 without compromising Bayesian evidence. We identify outstanding challenges interpretability, calibration of uncertainties under noisy data, hybrid modelling, and the generalization of models across instruments and planet populations and outline a research roadmap spanning the JWST era and beyond into Ariel's launch in 2029.
Muallim Yakubu, Vwavware Oruaode Jude
Department of Industrial Physics, Enugu State University of Science and Technology · Department of Physics, Dennis Osadebay University, Asaba, Delta State, Nigeria