# Known-object test, TOI-5696.03

> Drafted by `python -m cygnus.campaign report` from the runner's outputs; reviewed 2026-09-25.
> The escalation and reviewer notes below were added by hand; every other number is the runner's.

## Escalation (2026-09-25)

**Positive control failed on data that covers the epoch.** Sector 72 (`tess2023315124025-s0072-...`)
contains the catalogued epoch (BJD 2460284.7141) with 242 usable in-transit cadences, yet the screen
recovered only 820 ± 48 ppm against a catalogue depth of 1204 ppm — `not_recovered`. `docs/AGENT_RUNBOOK.md`
makes a positive-control failure on covered data an escalation. Per that rule, no threshold was tuned
and no state was changed; the failure is reported as-is.

Most likely cause, from the runner's own calibration: at sector 72's k* = 3 the 90 %-completeness depth
is 5000 ppm for an 8 h signal, and the catalogued transit is 1204 ppm over 8.09 h — about four times
below the detection floor. A shallow-transit-below-sensitivity failure is therefore the first
hypothesis, ahead of a wrong epoch or a wrong TIC. It cannot be distinguished further without an
independent reduction or a deeper light curve, neither of which is built here.

Two further notes for a reviewer: the TIC is a multi-signal host (TESS_TOI lists TOI-5696.01 PC,
.02 CP and .03 PC on the same TIC), so a blend with a neighbouring signal is a live alternative; and the
three persistent events all lie in sector 46, which does not cover the epoch.


- Campaign spec: `campaigns/toi-5696-03.yaml`
- Parent queue: `tess-mono-01`
- Ledger runs: calibrate_screen #285, fetch_products #284, known_signal_recovery #286, period_aliases #288, prior_art #289, residual_screen #287
- Runner finished (UTC): 2026-09-25T04:18:11Z

## Bottom line

Positive control **failed**: BJD 2460284.7141: not recovered, depth 820 ± 48 ppm (catalogue 1204 ppm).
Outside the catalogued epoch the screen left 14 threshold entries forming **5 distinct event(s)**, **3 persistent** (SAP and PDCSAP, two or more baselines). None is vetted; see *Screen events*.

This is a pipeline check on a known object, not a discovery claim. No period is implied by a single transit.


## Reviewer notes (2026-09-25, supersede pass)

**Superceded by the multi-archive runner:** re-run through `python -m cygnus.multi` on 2026-09-25, steps fetch_products, calibrate_screen, known_signal_recovery, residual_screen, period_aliases, prior_art (ledger fetch_products #1585, calibrate_screen #1586, known_signal_recovery #1587, residual_screen #1588, period_aliases #1589, prior_art #1590). Same retrieved products, same science outputs: entries, depths, k-star calibration and aliases match the production-runner results recorded here; the multi runner additionally records a screen-suitability census and a red-noise systematics model per light curve. The record is regenerated by the multi runner; nothing in the review changes.

## Reviewer notes (2026-09-25)

Positive control **failed** on data that covers the epoch, so this is escalated above. Sector 72
contains the catalogued epoch (BJD 2460284.7141) with 242 usable in-transit cadences, but the screen
found only 820 ± 48 ppm where the catalogue gives 1204 ppm — `not_recovered`. The other two retrieved
light curves (sectors 45, 46) do not cover the epoch.

The sensitivity table gives the most likely explanation: at sector 72's calibrated threshold (k* = 3) the
90 %-completeness depth is 5000 ppm for an 8 h signal, and the catalogued transit is 1204 ppm over
8.09 h — roughly four times shallower than the detection floor. So the failure is consistent with the
signal being below sensitivity rather than with a wrong epoch; it is recorded, not tuned around.

The three persistent screen events (2459563.12910 / .16521 / .21800, sector 46) span ≈ 0.09 d, i.e. one
contiguous ~0.25–0.27 % feature, not three transits. Sector 46 does not cover the catalogued epoch.
Catalogue cross-match answered by all four services: TESS_TOI lists TOI-5696.01 (PC), .02 (CP) and .03
(PC) on this TIC — a multi-signal TIC where the other dispositions matter for any interpretation.

## Target

| Field | Value | Source |
|---|---|---|
| name | TOI-5696.03 | NASA Exoplanet Archive TOI table (copied in the spec) |
| tic | 393546540 | NASA Exoplanet Archive TOI table (copied in the spec) |
| ra_deg | 161.156925 | NASA Exoplanet Archive TOI table (copied in the spec) |
| dec_deg | 6.333475 | NASA Exoplanet Archive TOI table (copied in the spec) |
| t0_bjd | 2460284.714061 | NASA Exoplanet Archive TOI table (copied in the spec) |
| depth_ppm | 1204.4442717 | NASA Exoplanet Archive TOI table (copied in the spec) |
| duration_h | 8.0912058 | NASA Exoplanet Archive TOI table (copied in the spec) |
| tmag | 9.1352 | NASA Exoplanet Archive TOI table (copied in the spec) |
| catalogue_row_updated | 2024-11-23 12:03:04 | NASA Exoplanet Archive TOI table (copied in the spec) |

## Products

| Product | Sector | Covers catalogued epoch | SHA-256 (first 16) | Retrieved now |
|---|---|---|---|---|
| `tess2023315124025-s0072-0000000393546540-0267-s_lc.fits` | 72 | True | `4a77c6806b7c1057` | True |
| `tess2021310001228-s0045-0000000393546540-0216-s_lc.fits` | 45 | False | `82aeabf2edcb94f7` | True |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 46 | False | `60b33114b9c3d990` | True |

## Positive control (catalogued transit)

| Product | Epoch (BJD) | State | Usable in-transit cadences | Measured depth (ppm) | Catalogue depth (ppm) | Entry offset (h) |
|---|---|---|---|---|---|---|
| `tess2023315124025-s0072-0000000393546540-0267-s_lc.fits` | 2460284.71406 | not_recovered | 242 | 820 ± 48 | 1204 | — |
| `tess2021310001228-s0045-0000000393546540-0216-s_lc.fits` | — | epoch not in this light curve | — | — | 1204 | — |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | — | epoch not in this light curve | — | — | 1204 | — |

Depth: median PDCSAP residual inside ±duration/2 about the catalogued epoch, against a 2-d running median; the error is statistical only. A depth unlike the catalogue's may reflect dilution, detrending or an epoch/duration error; it is reported, not used as a pass mark.

## Calibration and sensitivity

| Product | k* (sign-flip null) | At grid floor | 90 % completeness depth at k = 5, by duration | at k* |
|---|---|---|---|---|
| `tess2023315124025-s0072-0000000393546540-0267-s_lc.fits` | 3 | False | 1h: 5000, 2h: 5000, 4h: 5000, 8h: 5000 | 1h: 5000, 2h: 2000, 4h: 2000, 8h: 5000 |
| `tess2021310001228-s0045-0000000393546540-0216-s_lc.fits` | 3 | False | 1h: 5000, 2h: 5000, 4h: 5000, 8h: 5000 | 1h: 5000, 2h: 5000, 4h: 2000, 8h: 5000 |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 3 | False | 1h: 5000, 2h: 5000, 4h: 5000, 8h: 5000 | 1h: 5000, 2h: 5000, 4h: 2000, 8h: 5000 |

The null result (if any) excludes only dips deeper than the 90 %-completeness depth for their duration.

## Screen events outside the catalogued epoch

Entries merged where they overlap in time. *Persistent* = SAP and PDCSAP at two or more baselines.

| Product | Mid time (BJD) | Deepest median residual | Max cadences | Flux | Baselines (d) | Persistent |
|---|---|---|---|---|---|---|
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 2459563.12910 | -0.00269 | 2 | PDCSAP+SAP | 1, 2, 3 | yes |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 2459563.16521 | -0.00259 | 2 | PDCSAP+SAP | 2, 3 | yes |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 2459563.21800 | -0.00250 | 2 | PDCSAP+SAP | 2, 3 | yes |
| `tess2021336043614-s0046-0000000393546540-0217-s_lc.fits` | 2459563.04298 | -0.00236 | 2 | SAP | 3 | no |
| `tess2023315124025-s0072-0000000393546540-0267-s_lc.fits` | 2460272.20995 | -0.00230 | 2 | PDCSAP | 3 | no |

None has been vetted: centroids, pointing, background, momentum dumps and other reductions are **not tested**. Most screen events in TESS light curves are systematics; each is at most an unverified lead.

## Catalogue cross-match

**TOI-5696.03**

- NASA_Exoplanet_Archive (done, 2026-09-25): no match in NASA_Exoplanet_Archive within 30" as of 2026-09-25T04:18:04Z
- TESS_TOI (done, 2026-09-25): 3 match(es) in TESS_TOI within 30" as of 2026-09-25T04:18:06Z: TOI-5696.01 (TIC 393546540, disposition PC); TOI-5696.02 (TIC 393546540, disposition CP); TOI-5696.03 (TIC 393546540, disposition PC)
- VSX (done, 2026-09-25): no match in VSX within 30" as of 2026-09-25T04:18:08Z
- SIMBAD (done, 2026-09-25): 1 match(es) in SIMBAD within 30" as of 2026-09-25T04:18:09Z: BD+07  2351 (PM*)

## Checks

| Check | State | Note |
|---|---|---|
| Product integrity (SHA-256) | passed | 3 product(s) checksummed at first retrieval |
| Known-signal recovery (positive control) | failed | BJD 2460284.7141: not recovered, depth 820 ± 48 ppm (catalogue 1204 ppm) |
| Calibrated false-alarm threshold (sign-flip null) | passed | screen run at each light curve's own k* (3, 3, 3; ≤ 0 persistent null events outside the veto) |
| Synthetic signal injection–recovery | passed | completeness for the reference box (2000ppm_4h) at each light curve's k*: 90%, 90%, 90% (pass mark 90%); 90%-completeness depths are in calibration.json |
| Catalogue cross-match | passed | 1 target(s) × 4 services, radius 30″; 4 answered, 0 errored; results in the ledger prior_art table |
| Period aliases (repeat events) | not_tested | catalogued transit not recovered; no reference depth |
| Alternative detrending | not_tested |  |
| Difference-image centroids / blend audit | not_tested |  |
| Pointing / jitter correlation | not_tested |  |
| Literature (ADS) audit | not_tested |  |

## Reproduction

```bash
python -m cygnus.campaign run campaigns/toi-5696-03.yaml
python -m cygnus.campaign report campaigns/toi-5696-03.yaml
```

## Addendum 2026-09-26: suite-expansion checks

Re-run on 2026-09-26 with the steps added by the suite expansion (`docs/SUITE_EXPANSION.md` §7.3) and the corrected moving-object check (TESS-centred SkyBoT positions, distance-aware). The earlier science outputs (screen, calibration, positive control, aliases, cross-match) were compared leaf by leaf and are unchanged apart from timestamps. States and notes below are copied from `sky_record.json`; the review above is not changed by them unless a note says so.

| Check | State | Note |
|---|---|---|
| Target-to-Gaia identification (proper motion propagated) | passed | TOI-5696.03: Gaia DR3 3859284748342110208 at 0.00" (propagated 2016.0 → J2015.5; 0.03" unpropagated, proper-motion shift 0.03") |
| Stellar priors (Gaia colour and parallax) | passed | TOI-5696.03: Teff 5269 K, R* 0.81 ± 0.07, M* 0.88 ± 0.09, ρ* 1.64 ± 0.43 ρ☉ (dwarf sequence, M_G 5.55, no extinction) |
| Blend and dilution census (Gaia DR3 cone) | inconclusive | TOI-5696.03: 1 Gaia neighbour(s) within 52.5", contamination 0.33%; depth 1204 ppm (catalogue depth); 1 could produce it if fully eclipsed (brightest 3859283962363313152, 48.5", ΔG 6.21); a centroid test is needed |
| Pointing and quality census per event | passed | 3 persistent event(s), 3 clean; no artifact quality bit within ±0.25 d and no centroid, pointing or background shift beyond 5σ |
| Moving objects at screen-event epochs | inconclusive | 3 event epoch(s) queried in SkyBoT (observer C57, r=600"); no known object bright enough within 63" plus its motion at any queried epoch (supports, does not prove, a non-asteroid origin); 1 epoch(s) not answered |
| Variable-catalogue collision (VSX) | passed | TOI-5696.03: no VSX entry within 10" |
| Object-class guard (SIMBAD) | passed | TOI-5696.03: BD+07  2351 otype PM* (star_or_other) at 1.0" |
| Independent repetition (other MAST collections) | not_tested | no repeat candidate with allowed period aliases |
| Independent-epoch confirmation (ZTF) | not_tested | no repeat candidate with allowed period aliases |

`failed` on the per-event census means at least one persistent screen event carries an in-event artifact flag or pointing shift (`event_census.json`); it is a statement about those events, not about the catalogued signal.
