us-program-research（美国计划研究）

US Academic Program Research — Complete Workflow

Purpose

Run structured research on US academic programs with credential analysis, parallel discovery, adaptive scorecards, and generation of an actionable application plan in a deliverable format.

When to Use

Use this skill when the task requires:

• Selecting and ranking programs (PhD, Master's, or Bachelor's) in the US
• Detailed comparison of curriculum, costs, and admission requirements
• Application strategy based on profile, budget, and timeline
• A final consolidated document for decision-making and execution

Progress Tracking

Display progress before each research phase:

[███░░░░░░░░░░░░░░░░░] 15% — Phase 0: Program Type Detection & Input Collection
[██████░░░░░░░░░░░░░░] 30% — Phase 1: Profile & Credential Analysis
[█████████░░░░░░░░░░░] 45% — Phase 2: Parallel Discovery (4 subagents)
[████████████░░░░░░░░] 60% — Phase 3: Parallel Deep Research (4 subagents)
[███████████████░░░░░] 75% — Phase 4: Adaptive Scorecards & Ranking
[█████████████████░░░] 85% — Phase 5: Document Generation
[████████████████████] 100% — Phase 6: Inline Report & Recommendations

Workflow

Follow the phases defined below in sequence, maintaining source traceability and separating facts from inferences.

Personalized research on US academic programs (PhD, Master's MS/MBA/MPS, or Bachelor's). Analyzes the candidate's profile, runs parallel searches via subagents, identifies hidden gems, applies an adaptive scorecard, and generates a complete ACTION_PLAN.md with rankings, curricula, costs, and a step-by-step checklist.

Output language: Portuguese (matches user research context). Queries/subagents: English (required for search quality).

Execution Instructions

1. PHASE 0 — detect program type (FIRST QUESTION)
2. PHASE 0B — collect all inputs before researching
3. PHASE 2 — launch 4 subagents in ONE message (true parallelism)
4. PHASE 3 — launch 4 deep-research subagents in ONE message
5. PHASE 4 — apply adaptive scorecard after all research is complete
6. PHASE 5 — generate full document and save
7. PHASE 6 — present inline report to user

CRITICAL: All search queries and subagent prompts MUST be in English.

PHASE 0 — Program Type Detection

Use AskUserQuestion with: "What type of US academic program are you looking for?"

Options:

1. PhD / Doctorate — 4–7 years of research. Usually fully-funded (stipend + tuition).
2. Master's — MS, MBA, MPS, MEng (1.5–3 years). Executive, technical, online or in-person.
3. Bachelor's / Undergraduate — First or second degree (BS, BA). 2–4 years.

The entire workflow adapts to this answer.

PHASE 0B — Input Collection (2 rounds of AskUserQuestion)

Round A — Personal profile (all types, max 4 questions):

1. Full name + email
2. Country of origin + language(s) of diplomas
3. Highest current degree: institution, diploma type, GPA/grade (e.g., "8.1/10")
4. Desired field of study (e.g., Computer Science, Business Analytics, AI, Finance...)

Round B — Logistics and preferences (max 4 questions):

1. Target city/state in the US (or "open to any location")
2. Preferred format: in-person / 100% online / hybrid / no preference
3. Total budget: under $20k / $20–50k / $50–80k / over $80k / seeking funding
4. Desired start date: 2026 / 2027 / 2028 / flexible

Additional type-specific question:

• PhD: Specific research area? Known advisors? Funded programs only?
• Master's: Current role/level (Junior/Manager/Director/VP/C-Level)? Will you work during the program?
• Bachelor's: First or second degree? Transferable credits? Large/small campus preference?

Optional inputs (offer, do not require): CV, academic transcripts, prior executive courses, US immigration status (Green Card / F1-OPT / H1-B / other).

PHASE 1 — Profile & Credential Analysis

Determine foreign degree equivalency and evaluation strategy BEFORE scoring any program.

Action: Consult references/credential-analysis.md for:

• Degree equivalency table by country (includes risk for Brazilian Tecnólogo degree)
• WES vs ECE comparison by program type
• WES ICAP step-by-step logistics (8 steps + 2 scenarios: favorable vs unfavorable result)
• Anti-downgrading analysis for VP/C-Level candidates (Master's only)
• Advisor fit analysis for PhD + contact email template
• In-state tuition logic (Green Card + 12-month domicile = eligible)

Brazilian Tecnólogo = HIGH RISK. WES may evaluate as Associate's instead of Bachelor's. Recommend institutional letter + ECE as second opinion if WES result is unfavorable.

PHASE 2 — Parallel Discovery (4 Subagents)

CRITICAL: Launch all 4 subagents in ONE SINGLE message (Task tool × 4 simultaneous).

Divide research among 4 subagents with subagent_type="general-purpose":

Action: Consult references/subagent-prompts.md for the full prompts (in English) for each subagent, including all fields to collect and search queries.

After all 4 subagents complete:

• Deduplicate programs (same school may appear in multiple subagents)
• Verify accreditation: AACSB (business), ABET (engineering) — absence penalizes Brand Equity
• For PhD: verify if fully-funded vs self-funded
• Classify into groups: 1 (primary in-person), 1.5 (online brand equity), 2 (technical/deprioritized)
• Target: 12–20 unique programs for Phase 3

PHASE 3 — Parallel Deep Research (4 Subagents)

CRITICAL: Split the 12–20 programs into 4 groups (3–5 each) and launch in ONE single message.

For each program, collect: full curriculum, verified cost on official website, student reviews (Niche/Reddit/GMAT Club/GradCafe), admission requirements (GMAT/TOEFL/letters/deadline), alumni network.

Action: Consult references/subagent-prompts.md for the full deep research template (section "Deep Research Template") with all queries in English.

PHASE 4 — Adaptive Scorecards

Action: Consult references/scorecards.md for the complete scorecards.

Scorecard Selection by Program Type:

Master's — Adaptive Weights by Candidate Level:

*For VP/C-Level = "Executive Readiness". For non-executives = "Career Launch Potential".

Tier Classification:

💎 Hidden Gem = Score ≥ 70 AND Cost ≤ 50% of budget AND Satisfaction ≥ 8.0/10

PHASE 5 — Document Generation

Save as: {CANDIDATE_NAME}_US_PROGRAM_ACTION_PLAN.md (in current directory).

Action: Consult references/action-plan-template.md for the full document template in Portuguese, including all sections, ~30 aligned tables, and the Table Formatting Policy.

Required sections in the generated document: General Status → Priority Next Actions → Phase 1 (WES/ECE) → Phase 2 (Tests) → Phase 3 (Contact) → Phase 4 (Submission) → Evaluators → Selected Programs by Tier → Adaptive Scorecard + Full Ranking → Curriculum Comparison → Detailed Curricula → Cost Comparison → Admission Comparison → Why Each Program → Document Inventory → Notes.

PHASE 6 — Inline Report

Present to the user in chat (in Portuguese) after saving the file:

Block 1 — Candidate Profile (max 5 lines): type, field, role/education, estimated GPA, immigration status.

Block 2 — Top 5 Recommendations (compact table): | Rank | Program | Score | Cost | Highlight |

Block 3 — 💎 Hidden Gems (if identified): score ≥ 70 + low cost + satisfaction ≥ 8.0.

Block 4 — ⚠️ Alerts: programs without reviews (red flag), daytime-only for executives, advisors without recent publications (PhD), borderline degree → WES urgent.

Block 5 — URGENT Next Action: "What to do TODAY" — usually: start WES/ECE or contact advisor.

Block 6 — Generated File: ✅ Document saved: {NAME}_US_PROGRAM_ACTION_PLAN.md ({N} lines, {N} tables)

Additional Resources

Reference Files

Consult as needed during execution:

• references/credential-analysis.md — Degree equivalencies, full WES/ECE logistics, anti-downgrading, advisor fit (PhD), in-state tuition logic
• references/scorecards.md — Complete Scorecards A/B/C with detailed rubrics, ROI formula with numeric example, tier thresholds, and hidden gem definition
• references/subagent-prompts.md — Complete prompts (in English) for the 8 subagents: 4 discovery (Phases 2A/B/C/D) and deep research template (Phase 3)
• references/action-plan-template.md — Complete ACTION_PLAN.md template in Portuguese with all sections and aligned tables
• references/research-sources.md — Mandatory research sources and universal red flags by program type

Critical Rules

• All research queries and subagent prompts must be in English.
• Do not skip credential analysis before scoring/ranking.
• Do not present a final recommendation without citing sources and criteria.
• Clearly differentiate confirmed data, assumptions, and recommendations.

Error Handling

Example Usage

1. "Rank MS in Data Science programs in the US with a budget of USD 50k."
2. "Compare funded PhD options in Computer Science focused on distributed systems."
3. "Build an application plan for an MBA starting in 2027."