An anomaly-ranked catalogue of TESS sector 1 two-minute cadence light curves

An anomaly-ranked catalogue of the 12,905 targets retained from the two-minute-cadence input list of TESS sector 1, selected on the input-catalogue magnitude in the range 8 < Tmag < 13 and produced by the Atlas Robuste pipeline under protocol version 1.3.0 (registry seal sha256:0e6f8e37eb14f7bd64bf057477a22f8b). The input list comprised 12,999 distinct targets. The catalogue contains one row per retained target and includes 27 features extracted from the standardised light curve, an unsupervised anomaly score S, four robustness coefficients, the composite index PARS, an instrumental risk score, and a variability label where a counterpart was found within 10 arcsec in the TESS eclipsing binary catalogue, VSX or Gaia DR3 variability (1,638 counterparts, a base rate of 12.7%). The tmag column reports the magnitude recorded in the light-curve file header rather than the value on which the selection was made, and four targets fall marginally outside the nominal range on this column: TIC 219326123 (7.785), TIC 300871376 (7.855), TIC 260419879 (7.951) and TIC 277022714 (13.357). Users filtering on tmag should be aware of this. THE CATALOGUE IS RANKED, NOT CLEANED. Orbit edges were deliberately not masked, and instrumental artefacts are present at every level of the ranking. Users should read the accompanying provenance file before drawing population inferences from the catalogue. It records the complete selection function, including the 85 targets rejected for insufficient temporal coverage, the 4 with too short a baseline, the 3 without a matching data product, and 2 rejected during restandardisation, with their typed causes recorded in the provenance file. Components rather than a verdict. The score, each coefficient and the composite index are published side by side so that a user who disagrees with our weighting can recompose the ranking. Three of the four robustness coefficients were measured not to discriminate between injected astrophysical signals and injected instrumental artefacts, and are published for completeness only; this is the subject of the accompanying paper. Known limitations, each documented in the provenance file: 12.5% of targets have a periodogram peak pinned to an edge of the search grid, so their period is censored rather than measured; these targets are flagged by period_at_boundary. The factor-of-two degeneracy for equal eclipse depths is irreducible and unmarked, with odd_even_depth_ratio near unity indicating that the reported period may be half the true one. In addition, the two-minute-cadence sample is the mission's pre-selected target list, biased towards bright stars optimised for transit searches; occurrence rates derived from it inherit that selection. Files: the catalogue in Parquet and CSV, a column dictionary generated from the sealed feature registry, and a provenance record containing the protocol seal, the Git commit, dependency versions, per-file SHA-256 digests and the selection function.