Terminology
Canonical names for JobJitsu docs and UI. Prefer these terms; avoid synonyms in the same role.
Parent: PRODUCT_VISION.md · Brand chrome: ../brand/BRANDING_FOR_DEVELOPMENT.md
Product and UI
| Canonical term | Meaning | Avoid in the same role |
|---|---|---|
| Agent | Preparative career assistant (user-facing) | Bot, autopilot, chatbot (UI) |
| Agent · On-device | Status chrome for on-device intelligence | “Local LLM” in the status bar |
| Profile | One on-device identity (person) | Conflating with Path / résumé version |
| Path | Job-facing variant under a Profile (e.g. Fullstack, Mobile) | “Sub-profile”; separate accounts for every role |
| Queue | Approval holding area before send | Inbox (unless Sprint-1 shell label is explicitly noted) |
| Applications | Draft / tailor / track units of craft | Pipeline (in navigation) |
| Job Mail | Opt-in inbound mailbox sync + import feed (not a full inbox) | Mail, Inbox, Email as primary nav labels |
| Sources | Discovery / crawler entry (Coming soon until adapters ship) | Scraper as the user-facing label |
| Preferences | User rules, device folder, Agent model, appearance, Reset | Settings (prefer Preferences in product IA; Settings OK as shell label for the same surface) |
| Send | Explicit outbound egress | Auto-apply as the default story |
AI and knowledge
| Canonical term | Meaning | Avoid in the same role |
|---|---|---|
| AI Provider | Adapter that runs complete / embed / health | Mixing “runtime,” “backend,” and “provider” without definition |
| Model Manager | Load, select, unload, and monitor models for providers | Undocumented “model catalog” as a separate product |
| Local Intelligence | Product module for on-device AI | Using “LLM module” as the module name |
| Knowledge Base | On-device career knowledge store | Conflating with Timeline |
| Context Builder | Assembles minimal prompt context before inference | Dumping full history into every prompt |
| Resume Library | Stored resume versions and variants (optionally per Path) | Ad-hoc “résumé folder” naming in specs |
| Workflow | Declarative multi-step agent run | Free-form chat as the control plane |
| Task Queue | AI/work-unit scheduler (Pending → … → Cancelled) | Confusing with review Queue |
| AI Validation | Post-generation checks (formatting, ATS, skills, …) | Trusting first model output |
Discovery and integrations
| Canonical term | Meaning | Avoid in the same role |
|---|---|---|
| Job Provider | Source of job listings (discovery adapter) | Unspecified “scraper” / “board” without the Job Provider role |
| Application Pipeline | Application lifecycle stages (discovered → … → archived) | Using “Pipeline” as a primary nav label |
| Plugin | Capability-gated agent skill | Calling host UI/adapters “plugins” |
| Extension | Host contribution point (discovery, send, UI, …) | Calling agent skills “extensions” |
Privacy and egress
| Canonical term | Meaning | Notes |
|---|---|---|
| Local LLM | Technical name for an on-device language model | OK in advanced settings and architecture; not default status chrome |
| Trusted Automation | User-enabled reduced-approval mode (default off) | Never implied on by default |
| Timeline | Local audit / craft history | Distinct from Knowledge Base |
Document responsibility (short)
| Document | Owns |
|---|---|
| ../brand/PRODUCT_PHILOSOPHY.md | Philosophy only |
| PLATFORM_SPECIFICATION.md | Functional what |
| ../architecture/ | Structural how and architectural rules |
../../DEFINITION_OF_DONE.md + .cursor/rules/ | Engineering process |
| ../backlog/ | Build order |
| ../brand/ | Brand and UI nouns |
