SkySift candidates KIC 6699887: 15.114 and 47.691 d transit signals in Kepler photometry

Two dips, of 0.11 and 0.092 percent, repeat every 15.1 and 47.7 days on a star of magnitude 16.2. Ephemeris, 15.114 d candidate: P = 15.113696 ± 0.000054 d, T0 = BJD_TDB 2455662.9571 ± 0.0015 (epoch 46, the centre of the series), duration 4.02 +0.13/-0.11 h, depth 1,160 ± 50 ppm. Ephemeris, 47.691 d candidate: P = 47.69057 ± 0.00046 d, T0 = BJD_TDB 2455731.7581 ± 0.0041 (epoch 16, the centre of the series), duration 5.47 +0.10/-0.14 h, depth 993 ± 87 ppm. Reddening: E(B-V) 0.044 (16th to 84th percentile 0.041 to 0.065) from Bayestar19 (Green et al. 2019) at 841 pc (E(B-V) unchanged at the printed precision across the distance's 16th to 84th percentiles, 816 to 866 pc), a sightline average, adopted; through the Gaia DR3 law it gives E(BP-RP) 0.054 +0.025/-0.004, A_G 0.111 +0.052/-0.008 and A_Ks 0.016 +0.007/-0.001, and Teff 4,693 +192/-181 K (unreddened 4,582; full column 4,826), radius 0.730 +0.031/-0.024 Rsun (unreddened 0.712; full column 0.754) and planet radius 2.85 +0.19/-0.15 Earth radii (unreddened 2.77; full column 2.94); unreddened (E(B-V) = 0) is the earlier assumption; the 3D value is consistent with the full column (the full column, 0.113 ± 0.003 on the Schlafly and Finkbeiner 2011 scale, is an upper bound); the star sits about 149 pc above the plane, within the upper dust layer, so its true reddening may lie anywhere in the range from unreddened to the full column; the 3D value settles it; GSP-Phot E(BP-RP) 0.04, biased for cool dwarfs (Andrae et al. 2023), is not adopted. 15.114 d candidate. False-alarm probability: 1 in 290. 6 of 1,998 null searches (the inverted and scrambled light curves of 999 other KBonus stars, searched without the masked pass and alias resolution) reached the signal's SDE; across the 1,579-star sweep about 6 such peaks are expected by chance, and the vetting ladder, not the SDE, separates the candidate from them. False-positive probability: TRICERATOPS 1.1.0 gives 0.028 (0.012 without a planet on an unresolved bound companion), and 0.019 (0.0035 without it) with the 8 neighbours the difference images exclude removed, in none of its validation classes, which do not apply at 30-minute cadence (Giacalone et al. 2021); a four-scenario estimate without a bound-companion or diluted-planet scenario gives 0.0078, a lower bound. The pixel-excluded value (0.019, mainly STP) rests on our own evidence, the difference-image centroids; the all-neighbours value (0.028) rests on neighbour shares set by the PRF overlap rule within the catalogue's crowding budget, a bound rather than a measurement. 47.691 d candidate. False-alarm probability: 1 in 120. 16 of 1,998 null searches (the inverted and scrambled light curves of 999 other KBonus stars, searched without the masked pass and alias resolution) reached the signal's SDE; across the 1,579-star sweep about 13 such peaks are expected by chance, and the vetting ladder, not the SDE, separates the candidate from them. False-positive probability: TRICERATOPS 1.1.0 gives 0.037 (0.015 without a planet on an unresolved bound companion), and 0.024 (0.0029 without it) with the 8 neighbours the difference images exclude removed, in none of its validation classes, which do not apply at 30-minute cadence (Giacalone et al. 2021); a four-scenario estimate without a bound-companion or diluted-planet scenario gives 0.11, a lower bound. The pixel-excluded value (0.024, mainly STP) rests on our own evidence, the difference-image centroids; the all-neighbours value (0.037) rests on neighbour shares set by the PRF overlap rule within the catalogue's crowding budget, a bound rather than a measurement. This record is the submission pack of two transit candidates found by the SkySift engine in the KBonus Kepler Background light curves: for each, the detrended light curve, the ephemeris and parameters with their errors, the per-event depths, the vetting metrics, the figures and the standalone dossier, and for both a Research Note draft and the provenance of every query. What is not yet known: A season's box is cut by the stamp edge; A grazing or diluted eclipsing binary is not fully excluded by photometry alone; No spectrum: the planet radius depends on the stellar radius from Gaia colour and parallax; This is a candidate until follow-up confirms it. The two candidates were vetted singly; no joint fit was made.