/country-finder

Runs the full Country Finder pipeline. Collects your criteria, discovers candidate countries for remote hire and visa sponsorship as two separate tracks, generates per-country research prompts, ingests your results, scores each country against your requirements, and offers an optional reality check. Resumes from the last completed step if interrupted.

Flow

flowchart TD
  Start([User runs /country-finder]) --> Profile{profile.md<br/>exists?}
  Profile -->|no| Extract[Resume extraction —<br/>wait for upload and confirmation]
  Profile -->|yes| Sit
  Extract --> Sit

  Sit{situational-profile.md<br/>exists?} -->|no| SitQ[Collect location, citizenship,<br/>language — save to file]
  Sit -->|yes| State
  SitQ --> State

  State{.country-finder-state.json<br/>exists?} -->|yes| Resume[Inform user: resuming from step N]
  State -->|no| S1
  Resume --> S1

  S1[Step 1: Criteria intake<br/>remote and sponsorship tracks] --> S2
  S2[Step 2: Candidate discovery] --> S3
  S3[Step 3: Research prompt generator<br/>ready-to-copy prompts per country] --> S4
  S4[Step 4: Data ingestion<br/>one country at a time] --> S5
  S5[Step 5: Scoring<br/>deep-reasoner agent] --> RCQ

  RCQ{User wants<br/>reality check?} -->|no| SCQ
  RCQ -->|yes| S6[Step 6: Reality check<br/>deep-reasoner agent]
  S6 --> SCQ

  SCQ{User wants<br/>salary data?} -->|yes| SC[Offer /salary-calculator<br/>scoped to one country]
  SCQ -->|no| Done([Results delivered])
  SC --> Done

Steps

Profile check

Checks for profile.md in the workspace. If absent, reads prompts/shared/resume-extraction-prompt.md, waits for the user to upload their resume, and waits for explicit confirmation of the extracted profile before continuing. The profile is reused on all subsequent runs without re-extraction.

Situational profile

Checks for situational-profile.md. If absent, asks five questions: current location, citizenship, any known immigration friction tied to that citizenship, languages spoken, and required work language. Saves answers to situational-profile.md for reuse across sessions.

State check

Checks for .country-finder-state.json. If found, reads last_completed_step and informs the user which step will resume. If absent, creates the file with last_completed_step: 0 and starts from Step 1. Updates the file after each step completes.

Step 1 — Criteria intake

Collects hard requirements for both tracks separately. Remote track: minimum acceptable monthly salary (exact amount and currency) and maximum time zone difference (in hours). Sponsorship track: relocation openness, timeline, and dealbreakers. Exclusions: countries or regions to skip entirely. Vague answers such as “reasonable,” “flexible,” or “close” are rejected — exact numbers and clear yes/no answers are required before proceeding.

Step 2 — Candidate discovery

Generates a grounded list of candidate countries for each track, based on the criteria from Step 1. Remote and sponsorship candidates are listed separately.

Step 3 — Research prompt generator

Generates ready-to-copy research prompts for each candidate country on each track. The user copies these prompts and runs them in separate research sessions to gather real-world data. Claude does not generate the research itself.

Step 4 — Data ingestion

Accepts pasted research results one country at a time. Validates each message: one country per message, all required fields present, no silent overwrite if a country was already stored. Data is preserved verbatim — no analysis, scoring, or summarizing during ingestion.

Step 5 — Scoring

Claude asks whether to use the deep-reasoner subagent (Opus, high effort) for higher reasoning accuracy — if declined, the step runs with your current model. Scores each stored country against the criteria from Step 1, keeping remote hire and sponsorship tracks completely separate. Each country receives a fit classification (Strong / Moderate / Weak) and a confidence level (High / Medium / Low). Every excluded country requires a specific, evidence-based reason — vague dismissals are not accepted.

Step 6 — Reality check (optional)

Claude asks before running. If you confirm, Claude then asks whether to use the deep-reasoner subagent (Opus, high effort) — if declined, the step runs with your current model. Applies a deeper audit of the scoring output. If you decline the reality check entirely, the salary handoff offer appears immediately.

Handoff

After Step 5 scoring — or after Step 6 if you ran it — Claude asks whether you want salary data for any of the results and offers to run /salary-calculator scoped to a single named country.

Stop conditions

Profile not yet uploaded. Claude waits — it does not proceed or fill in placeholder data.
Any step instructs Claude to wait. Claude stops and waits. No guessing, no assumptions.
Vague answer to a criteria question. Claude asks again for an exact value before continuing.
Data ingestion receives multiple countries, missing fields, or a duplicate. Claude stops and explains the issue before storing anything.