import_url_extract

Status: Shipped — but only as a fallback.

What it does

Extracts job posting fields (title, company, description, etc.) from a page's markdown content using AI.

Why it exists

docs/marketing/application-tracking.md documents URL import as explicitly non-destructive: it returns parsed fields for review, and never auto-creates an application. This AI path is the fallback for postings without structured markup — see docs/superpowers/specs/2026-08-21-job-posting-jsonld-extraction-design.md.

How to use it

Web UI / extension: "Import from URL." The route tries extractJobPostingFromJsonLd (schema.org JobPosting parsing, no AI) first, and only falls through to this skill when the page has no usable JSON-LD.

What it reads

The fetched-or-caller-provided page content, converted to markdown and sliced to 8000 characters for the model. The application's persisted jobDescription field keeps the full untruncated markdown regardless of what the model saw.

What it will not do

It does not read your resume, Career Library, or repositories. It does not run at all when JSON-LD extraction already succeeded.

Source

apps/api/src/ai/skill-tools.ts:360 · apps/api/src/routes/applications.ts:1016 (POST /applications/import-url) · packages/mcp-tool-server/src/core/tools/applications.ts:93-108 (import-application-from-url)