Bug: Automation resume picker hides uploaded resumes

[frank-mendez]

Description

The automation wizard cannot select existing resumes that were uploaded as PDF or DOCX files. These resumes appear on the Profile page, but the automation resume picker reports that no resumes are available.

Steps to reproduce

1. Open Profile and create or import a resume with an attached PDF or DOCX file.
2. Leave the resume file-backed without completing the structured-section import, or use an existing file-backed resume.
3. Open Automations and click Create Automation.
4. Advance to Step 3: Select resume for matching.
5. Open the Resume for Matching dropdown.

Expected behavior

Existing resumes belonging to the user should be available for selection, or the wizard should provide a clear action to structure the uploaded resume before continuing. A file-backed resume should be usable for matching if the product accepts it as an existing/imported resume.

Actual behavior

The dropdown is empty and shows:

No resumes found. Please create a resume with enough content in your profile first.

The resume is visible in Profile and has a valid attached document, but it is excluded from the automation page because the page requests resumes with at least two structured ResumeSections. Uploaded files are stored before structured sections are created, so file-only resumes are filtered out.

Technical details / evidence

• src/app/dashboard/automations/page.tsx calls getResumeList with MIN_RESUME_SECTIONS_FOR_SELECTION.
• src/actions/profile/resume.ts filters the result using _count.ResumeSections >= minSections.
• The automation runner reads ContactInfo and ResumeSections for matching and does not read the attached PDF/DOCX file. Removing the picker filter alone would expose resumes that later produce an empty or unusable match input.
• On the local reproduction, five existing two-page PDF resumes were present, each with approximately 6k+ characters of extracted text, but all had ResumeSections = 0; the filtered eligible count was 0.

Environment

• JobSync v1.1.18
• Linux / Docker-backed SQLite environment
• Chrome 150

Suggested direction

Make the file-backed resume and automation matching flows use the same supported representation. Possible approaches are:

• structure/import the uploaded file before it becomes eligible for automation;
• have automation matching extract and use the attached PDF/DOCX text; or
• keep the structured-section requirement but show an actionable Structure with AI prompt/link from the automation wizard.

Related issues

• Resume too short for AI despite full file being available #76 — Resume too short for AI despite full file being available
• Job form Resume dropdown empty; empty resumeId can cause Prisma P2003 #91 — Job form Resume dropdown empty; empty resumeId can cause Prisma P2003

[frank-mendez]

Implemented in PR #110.\n\nThe automation picker now includes resumes with an attached PDF/DOCX even when structured sections have not been imported, and the runner extracts that document for matching. Structured resume content remains the preferred source when available.\n\nValidation: lint and TypeScript checks pass; focused regression tests pass (28 tests). The full suite passed 2,400 tests, with the existing backup round-trip setup blocked by a Prisma schema engine error.

[Gsync]

Thats intentional to filter out the empty resumes or not enough content, using pdf/doc files directly for llm call makes an impact on output quality and token usage. While it's true that workflow can be improved but thats not a priority for now.