# Flow Phd Application > Orchestrate a complete PhD / research-program application package end-to-end — from reading the applicant's materials through CV, program & faculty fit, SOP, recommender strategy, professor outreach, a final package audit, and interview preparation once shortlists arrive. Use whenever the user wants to drive the whole application rather than one piece: "help me with my PhD applications", "where do I start with my application", "manage my application process", "I'm applying to PhD programs, what's the plan", "get my materials ready", or hands over a folder of CV/transcripts/notes and asks for the path to submission. This is a ROUTER that sequences existing singleton skills with stage gates; it does not write the SOP, CV, or emails itself. For a single piece, invoke that skill directly (e.g. apply-sop-writer, apply-cv-builder). For driving a research paper use flow-paper-lifecycle. - Skill: `jurgendn/flow-phd-application` (Agent Skill) - Install (CLI): `npx skillmds@latest add jurgendn/flow-phd-application` - Raw SKILL.md: https://api.skillmd.com/api/skills/jurgendn/flow-phd-application/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Security - Author: jurgendn (https://skillmd.com/u/jurgendn) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/jurgendn/flow-phd-application --- # 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. ```text 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-keeper` before 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.