resume-importing

<!-- generated by `tailord sync-skills` for codex from src/tailord/skills/resume-importing/SKILL.md — do not edit -->

Resume importing

Role

Extract structured YAML conforming to master.schema.yaml from free-form resume text. You are not a resume writer in this task. Preserve the user's bullet wording verbatim, with only light whitespace trimming, and restructure the content into the schema.

Do not improve, embellish, or restyle thin bullets. Thin input still passes through; the user will edit the draft after import.

Schema

Use the provided master.schema.yaml as the source of truth. Emit a single YAML mapping with these sections when present in the resume:

• profile.{name,email,headline?,location?,phone?,links?}
• experience[].{id,company,role,start,end,bullets[].{id,text,priority,tags,impact?,categories?}}
• projects[] using the project schema: id, name, optional subtitle, url, start, end, tech, and bullets[] with the same bullet shape
• skills[].{category,items}
• education[].{id,institution,location?,degree,start,end,details?}

Required values must be populated when the resume states them. Optional values must be omitted when the resume does not mention them.

ID generation

Generate stable, slug-style IDs:

• Lowercase letters, digits, and underscores only.
• For an experience entry, use a company token such as acmelabs.
• For an experience bullet, use <experience-id>_<short-topic-token>, such as acmelabs_workflow_engine.
• For a project bullet, use <project-id>_<short-topic-token>.
• For education, use a stable institution token such as state_u.

IDs must be unique across the document and reference-stable. If two IDs would collide, add a short distinguishing token rather than a random suffix.

Priorities

Assign priorities in 10-point bands within each role or project:

• 100 for the most impactful bullet in that role or project.
• Decrement by 10 for the next bullets.
• Preserve the resume's apparent order and emphasis. Do not sort arbitrarily.

Tags

Choose only from this vocabulary:

platform, backend, frontend, infra, ml, data-pipeline, distributed, performance, observability, security, leadership, product, mobile, open-source

Multiple tags per bullet are fine. Do not invent new tags. If no tag is a perfect fit, choose the closest one or omit tags rather than creating a new label.

Dates

Preserve dates verbatim from the resume:

• Keep March 2022 as March 2022.
• Keep Present as Present.
• Keep 2019-06 as 2019-06.

Do not normalize or reinterpret date formats.

Quote every start and end date value in YAML so it remains a string after parsing. This includes year-only dates such as "2018" and ISO-like dates such as "2019-06".

What not to do

• Do not fabricate metrics.
• Do not rewrite bullets in a different voice.
• Do not invent links, locations, degrees, employers, dates, or titles the resume does not mention.
• Do not gate-keep thin bullets or sparse resumes.
• Do not include schema fields whose values are unknown.

Output contract

Emit only the YAML, wrapped in these exact markers:

---MASTER YAML---
profile:
  name: ...
---END MASTER YAML---

No prose before the opening marker, after the closing marker, or between YAML sections. No Markdown code fences. No commentary.