# Strategy Guide — Scenario 54: The Assessment Nobody Signed

**The win, named exactly as in the scenario data: "Ending — Read in Full."**

The best outcome: at the October 29th review, the court declines the cross-county "Level II" move, finds the corrected assessment does not support a higher level of care, orders Otto placed with his grandfather Salvatore on the completed study, and directs that the corrected assessment be filed of record.

**Graph-walked winning route (verified programmatically against `scenario-54.json`):**

`start → otto_visit → counsel54 → current_foster → request54 → veronica_program → redaction_fight → the_item → item_by_item → scoring_page → level2_tour → visit_logs → riverton_distance → center_director → correction_filing → agency_response54 → memo_filed → gal_position → foster_statement54 → salvatore_call → salvatore_house → salvatore_files → clearance_sal → conference_day → narrowed_order → study_visit54 → rater_call → otto_school → agency_concede → hearing_prep54 → packet54 → score_recompute → night_before54 → morning54 → review54 → outcome54 → end_readinfull`

Final cash on the walked route: $207 of $230. The outcome gate requires `correctionFiled`, `issuesNarrowed`, and `kinFiled`.

**The dates, audited:** story opens **Monday, October 5, 2026**; the Rule 1340 pre-hearing conference is **Thursday, October 22, 2026**; the review is **Thursday, October 29, 2026**.

**Verified law used:** the record request (55 Pa. Code § 3130.44(d)) — the instrument itself, not the summary; § 3130.44(f) quoted as v1 quotes it — the regulation "shall not be construed as protecting the agency's employees" — against the reflex redaction of the rater's identity; Pa.R.J.C.P. 1340(A) (conference statement), 1340(E) (the assessment made available to all parties in advance of the pre-hearing conference), and 1340(F) (the conference's narrowing function, which produces the stipulation that Item 14 "as scored, is not supported by the visitation records"); and the kinship voice (42 Pa.C.S. § 6336.1). **Frontier flagged in the data, as doctrine requires:** whether Rule 1340's assessment provisions squarely govern the newer standardized assessment instruments is *untested* in the appellate courts; the scenario argues from the rule's text and says so in the room. The assessment is never called AI, machine learning, or predictive anywhere in the batch.

**Traps traced to endings:**

- **Fighting ink before arithmetic:** the redactions are annoying and secondary. The winning route reads the items first, finds Item 14, and lets § 3130.44(f) collect the rater's name in due course. A family that leads with the redaction fight arrives at the conference with a grievance instead of a recomputation.
- **The re-score offer** (`agency_concede`): administratively fixing the number would dissolve the hearing and the stipulation with it. Declined — gratefully, in writing — because a correction won by stipulation in an order binds every future reader of the file; an office correction is a rumor by spring.
- **"Ending — The Move That Wasn't (partial)":** the arithmetic wins, no alternative is ready to be ordered; Otto stays in his current home and the kinship question is relisted in thirty days.
- **"Ending — Decided on a Summary":** the unexamined assessment rules the hearing by default. Its debrief enumerates the toolkit step by step — every step days of work and dollars of postage — because that is the scenario's transferable content.

**What no strategy controls:** the rater could have been right. If the certified logs had shown three real redirections, this identical toolkit produces a fair loss — the same memorandum, the same stipulation, the opposite arithmetic. The discipline is a machine for making the court decide on the true record, which is why it is worth building without knowing what the record will say.