create_resume_draft
ChatGPTCreate a resume on Interview Trix from extracted or collected facts, or retarget/merge an existing one. Set creationMode explicitly: fresh (no sourceResumeId; never silently reuse employment from chat context), retarget (sourceResumeId + jobDescription; facts from saved resume only), or merge_attachment (sourceResumeId + extracted payload overlay). templateId may be a display name (Classic, Harvard, Meridian) or an id. After create, offer score_resume_ats. For small fixes on the same resume, use update_resume_draft instead.
get_entitlements
ChatGPTReturn the user's Interview Trix plan, remaining interview credits, and whether AI mock interviews, ATS, and detailed reports are allowed. Call this before creating an interview.
get_interview_report
ChatGPTFetch the detailed report for a completed interview. Requires the detailedInterviewReport entitlement.
get_resume
ChatGPTFetch one resume's full structured content (header, summary, experience, education, skills, projects, and other sections) by resumeId. list_resumes only returns titles — you must call this before Q&A, rewriting, or retargeting a saved resume. Do not invent experience if this succeeds.
get_resume_pdf
ChatGPTReturn a short-lived download link for a resume PDF the user already generated on Interview Trix. Pass resumeId or name (name ties use the newest resume). Use get_resume when ChatGPT should read the resume content. If the user needs the file itself, they can open the link or download and attach it in chat. Does not generate a new PDF.
import_resume_text
ChatGPTExtract structured sections from plain resume text (after the host extracts PDF text in chat). Returns a preview and suggested creationMode. Then call create_resume_draft with mapped fields. Use merge_attachment + sourceResumeId when enhancing a saved resume with a new attachment.
initialize_mock_interview
ChatGPTCreate a draft AI mock interview and return a deep link to start on Interview Trix. Call this ONLY after the user picked a duration (via start_interview_flow or by saying 15/30) — never default to 30 or choose the duration yourself. Always pass resumeId from the resume created or listed in this chat when available. Use 30 only for Tech Pro/Enterprise. Omit department/discipline unless the user gave an exact allowed value.
list_interviews
ChatGPTList the user's recent mock interviews (status, role, score when completed).
list_resume_templates
ChatGPTCall this to show Interview Trix resume templates (17 names, ids, and thumbnail URLs). ChatGPT may render a thumbnail picker. Never pick a template yourself. Never ask the user to type a template id. After they pick a name (Classic, Harvard, Meridian) or tap the picker, immediately call create_resume_draft with that name — do not only acknowledge, and do not tell them to use a picker that is not visible.
list_resumes
ChatGPTList the user's Interview Trix resumes (id, title, ATS score, whether a PDF is ready) with editor links. This is metadata only — call get_resume with a resumeId to read jobs, education, skills, and other structured content before rewriting or retargeting.
score_resume_ats
ChatGPTRun ATS scoring on an existing resume. Optionally pass a job description for tailored scoring. Returns score, top issues, and an ATS editor link.
start_interview_flow
ChatGPTMandatory first step when the user wants a mock interview and has not stated a duration. Returns duration options (15 / 30 minutes), whether 30 is allowed on their plan, and language. The user must pick a duration — never choose one yourself and never default to 30. On ChatGPT a duration picker widget appears; on other clients ask one short question (15 or 30). After they pick, call initialize_mock_interview with that duration.
start_resume_flow
ChatGPTMandatory first step when the user wants to create or build a resume. Returns path choices (retarget saved, brand-new, upload file, merge attachment) and saved resume ids. Ask ONE path before template or intake. On ChatGPT a path picker widget may appear; on Claude/Gemini use the markdown buttons.
start_resume_intake
ChatGPTStart section-by-section resume intake (header, summary, current role, education, skills, then optional past roles, projects, certificates). REQUIRES a chosen templateId (name or id): if the user has not picked a template yet, call list_resume_templates FIRST and wait for their pick — calling this without a valid templateId returns the template picker (needsTemplate), not the intake form. Ask only the next section from the JSON, never the full list. LinkedIn and GitHub are optional — do not block on them. Pass prefill when header was collected earlier in this chat.
update_resume_draft
ChatGPTUpdate an existing resume in place (same resumeId). Use for header/title/subtitle fixes, typos, one bullet, or clearing a wrong portfolio field — do not recreate. Does not run full JD retarget (use create_resume_draft with sourceResumeId for that). After update, offer score_resume_ats.