A transit search where a panel of independent reviewers marks dips in blind-coded light curves, consensus is skill-weighted, and every candidate keeps its provenance. Built to follow the Planet Hunters CHEOPS workflow.
Probes only; Sentinels not run on this example. Sentinels run on Claude (Anthropic) by default; other providers are supported and labelled. Probes are deterministic matched-filter detectors.
Thresholds, weights and every input are written to manifest.json, so a result can be rerun exactly.
Passed candidates are the catalogue. Rejected and flagged events are kept on purpose, so a reader can see what the vetting decided and why.
Every reviewer's answer for every blind-coded subject. Disagreement is kept visible, because it is where the consensus is earned.
| reviewer | label | marks (BTJD) | rationale |
|---|---|---|---|
| volunteer-sensitive | potential planet | 1425.689–1425.776, 1430.456–1430.545, 1431.671–1431.762, 1435.180–1435.282 | simulated detector |
| volunteer-typical | potential planet | 1425.674–1425.759, 1431.698–1431.777 | simulated detector |
| volunteer-cautious | potential planet | 1425.685–1425.771, 1431.693–1431.771 | simulated detector |
| reviewer | label | marks (BTJD) | rationale |
|---|---|---|---|
| volunteer-sensitive | potential planet | 1414.510–1414.597 | simulated detector |
| volunteer-typical | potential planet | 1414.514–1414.596 | simulated detector |
| volunteer-cautious | potential planet | 1414.496–1414.596 | simulated detector |
*Small planets near the limit.* A bright K dwarf with two small known planets: b (about 1.6 Earth radii, 4.8 days, roughly 350 ppm deep) and c (17.2 days).
Run runs/tess-hd-15337-s4 (TESS SPOC 2-minute, sector 4; config configs/showcase/tess-hd-15337-s4.toml, sha256 e21c63a794b2aeb3…; code v0.1.0-6-g2bf0522). Light curves: public TESS data from MAST tess2018292075959-s0004-0000000120896927-0124-s_lc.fits (sha256 a0d03d3e07a18ab1…).
**Probes only; Sentinels not run on this example.** 3 dips (2 candidates, 1 flagged).
A small planet whose single transits are near the noise of one sector: it shows what the panel and vetting do at the edge of what a 10-minute-binned search can see, and the page says what it missed.
Published values: NASA Exoplanet Archive, table ps (every row for each planet), fetched 2026-10-22 (docs/results/showcase_reference.json). Each planet's transits are predicted from the archive row named below; SetuHQ's values are the medians over the dips on that ephemeris that were not rejected (the limb-darkened fit's Rp/R★, T14 and mid-transit depth, and the BLS period). Depth definitions differ between papers and this fit, so compare Rp/R★ and T14 first. A blank published value is one the archive does not hold for that paper; it is not compared.
Ephemeris from Rosário et al. 2024; compared with Gandolfi et al. 2019 (https://ui.adsabs.harvard.edu/abs/2019ApJ...876L..24G/abstract).
| quantity | SetuHQ | Gandolfi et al. 2019 | ratio | archive default (Rosário et al. 2024) | ratio |
|---|---|---|---|---|---|
| P (d) | 5.97872 | 4.75615 | 1.257 | 4.75598 | 1.257 |
| Rp/R★ | 0.0246 | 0.0176 | 1.397 | 0.0190 | 1.296 |
| T14 (h) | 2.36 | 2.49 | 0.949 | ||
| depth (ppm) | 536 |
The BLS period stored on these dips (5.9787 d) is not this planet's: within a 13-day subject the search paired dips that belong to different transits (another planet's, or an alias), so the period ratio is a miss, not a recovery, and the fit that used it (Rp/R★, depth) is less reliable than its T14. The dips sit on the published ephemeris; the period does not.
Transits predicted in the stored light curve: 5; with a dip on them: 1 (1 dip; 1 passed; mean offset -3.2 min).
Ephemeris from Rosário et al. 2024; compared with Gandolfi et al. 2019 (https://ui.adsabs.harvard.edu/abs/2019ApJ...876L..24G/abstract).
| quantity | SetuHQ | Gandolfi et al. 2019 | ratio | archive default (Rosário et al. 2024) | ratio |
|---|---|---|---|---|---|
| P (d) | 4.0429, 5.9787 (one per dip, BLS; not one period) | 17.17840 | 17.18055 | ||
| Rp/R★ | 0.0238 | 0.0256 | 0.931 | 0.0271 | 0.880 |
| T14 (h) | 2.27 | 2.25 | 1.007 | ||
| depth (ppm) | 588 |
The BLS period stored on these dips (4.0429, 5.9787 d) is not this planet's: within a 13-day subject the search paired dips that belong to different transits (another planet's, or an alias), so the period ratio is a miss, not a recovery, and the fit that used it (Rp/R★, depth) is less reliable than its T14. The dips sit on the published ephemeris; the period does not.
Transits predicted in the stored light curve: 2; with a dip on them: 2 (2 dips; 1 passed, 1 flagged; mean offset +0.7 min).
A known planet found again is a check of the pipeline, not a discovery. HD 15337 b: 1 of 5 transits (1 candidate; BLS period not the planet's); HD 15337 c: 2 of 2 transits (1 candidate, 1 flagged; BLS period not the planet's).
Panel: volunteer-sensitive, volunteer-typical, volunteer-cautious
Weights: volunteer-cautious=0.633, volunteer-sensitive=0.667, volunteer-typical=0.667
Completeness is the fraction of injected transits recovered as a vetted candidate, with the consensus centred within half a transit duration of the injection.
| depth (ppm) | duration (h) | subjects | recovered | completeness |
|---|---|---|---|---|
| 300 | 1.0 | 20 | 0 | 0.00 |
| 300 | 2.0 | 20 | 0 | 0.00 |
| 300 | 4.0 | 20 | 0 | 0.00 |
| 1000 | 1.0 | 20 | 3 | 0.15 |
| 1000 | 2.0 | 20 | 16 | 0.80 |
| 1000 | 4.0 | 20 | 20 | 1.00 |
| 3000 | 1.0 | 20 | 20 | 1.00 |
| 3000 | 2.0 | 20 | 19 | 0.95 |
| 3000 | 4.0 | 20 | 20 | 1.00 |
False positives: 0.000 vetted candidates per empty subject (40 empty subjects).
| id | kind | weight |
|---|---|---|
| volunteer-sensitive | simulated | 0.600 |
| volunteer-typical | simulated | 0.600 |
| volunteer-cautious | simulated | 0.400 |
Skipped (unavailable here): none
e21c63a794b2aeb3…{
"setuhq_version": "0.1.0",
"created_at": "2026-10-22T18:48:34+00:00",
"python": "3.13.16",
"platform": "Linux-6.18.44-fc-v140-x86_64-with-glibc2.39",
"packages": {
"numpy": "2.5.4",
"scipy": "1.18.1",
"astropy": "8.0.2",
"matplotlib": "3.11.2",
"lightkurve": "2.6.0"
},
"config_name": "tess-hd-15337-s4",
"config_sha256": "e21c63a794b2aeb3199eccccc7a36917d3cd23c0f937f756866e11b968abdf71",
"subject_days": 13.0,
"search_scope": "subject",
"claude_prompt_version": "2026-10-16.1",
"sentinel_prompt_version": "2026-10-16.1",
"inputs": [
{
"source": "MAST",
"author": "SPOC",
"sector": 4,
"exptime_s": 120,
"flux_column": "pdcsap_flux",
"time_unit": "BTJD",
"lightkurve_version": "2.6.0",
"file": "tess2018292075959-s0004-0000000120896927-0124-s_lc.fits",
"file_sha256": "a0d03d3e07a18ab133fe1fe31b9f3d45870a8c0f7a00d8dd2e5759bfd2e70f5e",
"crowdsap": 0.99993479,
"flfrcsap": 0.94710827,
"target": "TIC 120896927",
"mission": "TESS"
}
],
"reviewers": [
{
"id": "volunteer-sensitive",
"kind": "simulated",
"weight": 0.6,
"calibration": "calibrated",
"calibration_note": "measured by injection"
},
{
"id": "volunteer-typical",
"kind": "simulated",
"weight": 0.6,
"calibration": "calibrated",
"calibration_note": "measured by injection"
},
{
"id": "volunteer-cautious",
"kind": "simulated",
"weight": 0.4,
"calibration": "calibrated",
"calibration_note": "measured by injection"
}
],
"skipped_reviewers": [],
"skipped_reviewers_detail": [],
"weights": {
"volunteer-cautious": 0.4,
"volunteer-sensitive": 0.6,
"volunteer-typical": 0.6
},
"weights_source": "docs/results/reviewer_weights_full_panel.json (Probes only)",
"panel": {
"mode": "probes_only",
"sentinels_configured": [],
"note": "Probes only; Sentinels not run on this example"
},
"code": "v0.1.0-6-g2bf0522",
"counts": {
"light_curves": 1,
"subjects": 2,
"classifications": 6,
"consensus_events": 3,
"passed_vetting": 3
}
}