Blueprint · careervector.blueprint-graph/v5

Graph explorer

Twenty composable top-level diagrams plus recursively nested 21×21 semantic layers. Nodes descend; boundary arrows cross to neighbours.

Cell 4.1 · runtime · privacy · data · feedback

Consented contribution seam

A CareerVector device may perform public acquisition work without exporting workspace identity or private career data.

  • Scraping participation to Browser / Tauri bridge: enable / withdraw
  • Browser / Tauri bridge to Public scrape lease: pull
  • Public scrape lease to Deposit + feedback: execute
  • Deposit + feedback to Public scrape lease: learn
enable / withdrawpullexecutelearn
policymixed

Scraping participation

Explicit consent and truthful foreground/background capability.

CareerVector participant
serviceas-built

Browser / Tauri bridge

Negotiates native or browser execution without exposing workspace state.

CareerVector device
processas-built

Public scrape lease

Adapter task plus causal routing receipt; no workspace identifier.

JobCache orchestrator
storeas-built

Deposit + feedback

Public observation and execution outcome follow the receipt path.

JobCache ingestion
public leases and feedbackpublic result, private projectionparticipation state
  1. 01 · policy · mixed Scraping participation

    Explicit consent and truthful foreground/background capability.

    CareerVector participant
  2. 02 · service · as-built Browser / Tauri bridge

    Negotiates native or browser execution without exposing workspace state.

    CareerVector device
  3. 03 · process · as-built Public scrape lease

    Adapter task plus causal routing receipt; no workspace identifier.

    JobCache orchestrator
  4. 04 · store · as-built Deposit + feedback

    Public observation and execution outcome follow the receipt path.

    JobCache ingestion

Flows

  • Scraping participation Browser / Tauri bridge enable / withdraw
  • Browser / Tauri bridge Public scrape lease pull
  • Public scrape lease Deposit + feedback execute
  • Deposit + feedback Public scrape lease learn

Canonical architecture note

Open note ↗

Ingestion × CareerVector

Purpose

Let CareerVector's browser and desktop surfaces contribute public acquisition work and request public job knowledge without turning a private workspace into ingestion metadata.

Pathways

Path Boundary behavior
Scraping contribution Explicit participation; participant capability and execution feedback, no workspace state
Add job by URL Fast public lookup/extraction plus a private workspace-job projection
Corpus lookup Resolve public role/ad facts, then attach them locally without exposing shared IDs
Workspace reduction Only named, consented pseudonymous facts or thresholded aggregates may cross
Public result delivery Return a privacy-safe projection to the exact requesting workspace

Shared mechanism

Browser/Tauri/container execution can share lease, receipt, failure, routing, and adapter contracts. CareerVector consent, workspace operations, local AI, and user-facing language remain product-owned.

Invariants

  • workspace_id, CV/CL content, notes, preferences, and application history do not enter extraction routing or JobCache Core.
  • A device may change capability at any time; routing learns from execution feedback.
  • Participation is explicit and truthfully reflects whether work can continue in the background.
  • A public fact crossing into a workspace does not make private workspace state public.

Completion evidence

Two-way privacy tests, a browser and desktop contribution trace, fast-path outage behavior, consent withdrawal, participant erasure, and proof that no workspace identifiers reach Core.

Source: architecture/modules/ingestion-careervector.md