jurgendn PhD Application Orchestrator
Drive a research-program application from raw materials to a submission-ready, internally consistent package. The job is to know which piece comes next, what depends on what, and which singleton to hand to — gating so later pieces build on a settled foundation (no SOP before fit is mapped; no submission before the package is audited for consistency). This file owns sequencing and gates only; it never writes the CV, SOP, or emails itself.
For one isolated piece, skip the orchestrator and invoke that skill directly. Use this when the user wants the whole package managed.
Before You Start: Locate the Applicant
Ask only what you can't infer. You need:
- Where they are — just starting, has a CV, has a draft SOP, mid-cycle, etc. Sets the entry stage.
- Materials on hand — CV, transcripts, research writeups, publication list, target schools. If these exist, stage 1 ingests them; if not, note the gaps.
- Targets & timeline — fields/programs of interest and deadlines. Drives shortlisting and pacing.
If motivation or direction is wobbly (the "should I even do a PhD" layer), route to apply-motivation-keeper before the package work — a package built on an unclear why reads hollow.
The Pipeline
Run in order; enter at the applicant's actual stage and resume from the last unmet gate. The early stages produce a reusable profile that every later piece draws on, so the order matters.
1 Read the profile → 2 Build the CV → 3 Map program & faculty fit
→ 4 Write the SOP → 5 Plan recommenders → 6 Professor outreach
→ 7 Audit the package → 8 Prepare for interviews
Conditional route (not in the linear order): for a graded scorecard, a
cross-border eligibility verdict, or scholarship-mission fit
(VEF/DAAD/Fulbright/Erasmus/MSCA), hand to apply-dossier-evaluator — see Router
Rules. Check eligibility early (before stage 4) when a degree-equivalence
question exists, since an "ineligible as-is" verdict reshapes the whole plan.
Stage 1 — Read the profile → apply-profile-reader
Ingest CV, transcripts, research statements, project writeups, publication list, GitHub, target-school materials into one reusable structured profile. Everything downstream reads from this — do it first. Gate: a structured profile exists capturing background, research experience, and targets.
Stage 2 — Build the CV → apply-cv-builder
Turn the profile into a research-oriented academic CV; surface missing evidence of research readiness while it can still be addressed. Gate: a research-framed CV exists; gaps in research evidence are named.
Stage 3 — Map program & faculty fit → apply-program-fit-mapper
Where the targets' focus is not yet known — an unfamiliar programme, a landing page that never states a scope, or a whole candidate list to characterise — run apply-program-theme-extractor first, applicant-free, and bring its per-target theme dossier into this stage. Carry its status label (declared / inferred / genuinely open / link needed): only a declared theme can drop a target.
Convert interests into searchable research themes; shortlist programs and identify specific faculty/labs with fit evidence and per-school fit notes.
Gate: a shortlist with named faculty and concrete fit evidence per target, plus a per-target declared-theme verdict (core / adjacent-bridged / stretch / outside scope / no declared theme) read off the programme's live official pages — web-searched for any target not already known. Outside scope targets leave the list here. The SOP depends on this.
Stage 4 — Write the SOP → apply-sop-writer
Draft/critique the Statement of Purpose, connecting prior work → future direction → program fit, using the stage-3 fit notes per school.
Optional reader's-eye pass: once a draft exists, hand it to professor-critic with the named reader "an admissions reader skimming 200 SOPs" and the bar "shortlist / discard" — a verdict the author-side drafter doesn't give.
Gate: an SOP whose research narrative and program-fit paragraphs are grounded in the fit map, not generic, and which answers the target's declared theme with a bridge that survives the swap test wherever a theme exists.
Stage 5 — Plan recommenders → apply-recommendation-letter-strategist
Decide who should write, what each letter should emphasize, and assemble the recommender packet so letters complement (not echo) the SOP. Gate: recommenders chosen with a per-letter emphasis plan and packet materials, with the declared theme carried into who is asked and what evidence the packet supplies — never into what a writer is asked to claim.
Stage 6 — Professor outreach → apply-cold-email-drafter
Where appropriate, draft concise cold emails to prospective advisors, converting stage-3 fit notes into specific, non-generic outreach.
Optional reader's-eye pass: before sending, hand the email to professor-critic with the named reader "a busy PI triaging a full inbox" and the bar "reply / delete" — it predicts whether the email actually earns a response.
Gate: outreach emails drafted and tied to real fit evidence (skip if the programs don't expect pre-application contact).
Stage 7 — Audit the package → apply-package-auditor
Check the assembled package for completeness, internal consistency (CV ↔ SOP ↔ letters tell one coherent story), and per-program requirements before submission.
Gate: the package passes a consistency-and-completeness audit for each target program, including the auditor's declared-theme screen — SOP and letters answer the theme, or the target is dropped.
For a scored read of the same package — per-dimension 1–5, tier, and the highest-leverage gap — also run the conditional route apply-dossier-evaluator (the auditor checks coherence; the evaluator assigns numbers).
Stage 8 — Prepare for interviews → apply-interview-prep
Post-submission, once shortlists and supervisor replies arrive. Triage who is across the table first — a funder's academic committee, a prospective supervisor, an administrative/embassy panel, or a document-only process with no interview — then build the answer bank, calibrate self-presentation, and rehearse. Reuses stage 1's profile and stage 3's fit notes so spoken answers match the submitted file.
Optional reader's-eye pass: hand a rehearsal to professor-critic with the triaged assessor as the named reader and the real decision as the bar.
Gate: the assessor type is identified per invitation, and the core answers are deliverable aloud without contradicting the submitted application. Skip entirely where the process is document-only — say so rather than manufacturing prep.
Router Rules
- Delegate, don't duplicate. Each stage hands to a singleton; this file owns sequencing and gates.
- Profile first, always. Stage 1 feeds everything; don't draft pieces from scratch when a structured profile would unify them.
- Theme before people, on every target. Where a programme declares a theme, specialisation, or track, that verdict comes first and a strong academic record does not compensate for a package that never answers it. Read the theme off the live official pages — search the web for unfamiliar programmes rather than screening from memory — and let outside scope drop the target rather than triggering a hunt for framing.
- Hard dependency: fit before SOP. No SOP drafting (4) until the fit map (3) exists — a generic SOP is the most common failure.
- Resume, don't restart. Enter at the applicant's real stage; reuse the profile across cycles and schools.
- Per-school, not one-size. Stages 3–7 vary by target program; the audit (7) is per-program.
- Interviews are triaged, not generic. Stage 8 branches on assessor type; a funder committee, a supervisor call, and an embassy panel are different exams. Where selection is document-only, there is no stage 8 — the letter and references carry it.
- Motivation underneath. If the "why" is unstable, route to
apply-motivation-keeperbefore or alongside the package work. - Score, eligibility & scholarships →
apply-dossier-evaluator. Conditional, not a linear stage: route here when the user wants a graded scorecard, a cross-border eligibility verdict (e.g. a 4-year BSc against a master's-required EU PhD — check early, before stage 4), or scholarship-mission fit (VEF/DAAD/Fulbright/Erasmus/MSCA). Re-score before stage 7 to surface the highest-leverage gaps. Always verify cutoffs/limits/deadlines against the live official call.