initmahesh/mlai-community-labs/skill

Job Strategy Doc Builder

Turns a scattered "I want an AI job" goal into a structured strategy document, modeled on a diagnosis → guiding policy → objectives → action plan → roadmap → risk framework (the same shape strategy consultants use for corporate strategy, applied to one person's career). The point isn't to produce generic career advice — it's to force specific, personal answers out of the user via targeted questions, then organize those answers into something they'll actually reread and act on.

Two deliverables come out of every run:

1. A Markdown strategy doc (their reference copy, easy to edit later)
2. A self-contained HTML artifact (a visual version — sections as cards, a milestone timeline,

a KPI tracker) they can open in a browser and glance at without wading through prose

After the doc is built, offer interview-prep follow-ups — see "Follow-up menu" below.

Step 1 — Get the resume (and check for an existing strategy doc)

Before asking anything else, check whether the user already attached a resume/CV in this conversation or uploaded one previously. If yes, read it — it's your best source for the Diagnosis section (experience, domain, project history) and you should pull from it rather than re-asking things it already answers.

If no resume is present, ask for one before proceeding — the strategy doc is much weaker without it. Use AskUserQuestion:

• Question: "I'll build your strategy doc around your background. Do you have a resume handy?"
• Options: "I'll upload it now" / "Use my LinkedIn profile instead (I'll paste the URL/text)" /

"I don't have one ready — build from what I tell you"

If they choose the third option, proceed, but lean harder on the interview questions in Step 2 to fill the gaps a resume would normally cover.

Also check if a strategy doc from this skill already exists earlier in the conversation (look for a prior *-job-strategy.md file or a completed workshop). If one exists and the user's current request is really about interview prep (not building a new strategy), skip straight to "Follow-up menu" below and use that existing doc + resume as context.

Step 2 — Run the workshop (AskUserQuestion, in batches)

This is the core of the skill: don't silently invent the user's answers. The whole value of this exercise is forcing them to commit to specifics they'd otherwise leave vague. Use AskUserQuestion in three batches (max 4 questions per call). Give each question 3-4 concrete preset options plus room for "Other" — presets make it fast to answer, "Other" keeps it honest when none of the presets fit.

Batch A — Diagnosis (current state, challenges, opportunities):

1. Experience level / current domain (options spanning junior/adjacent → senior/leadership,

grounded in what the resume shows if you have one)

1. Employment status right now (employed full-time / between roles / student or new grad /

freelance)

1. Biggest challenge to landing this role (location constraints / employment gap or irrelevant

experience / skills-domain gap / limited time)

1. Biggest edge or unique asset (deep domain expertise / engineering-technical background /

strong network / a specific project or body of work)

Batch B — Guiding Policy & Objectives:

1. Overall approach they want to take (domain-expertise-first / portfolio-and-build-in-public /

certifications-and-network / direct outreach and referrals)

1. Non-negotiables — multiSelect (compensation must increase / location is fixed / must be

people-management track / must be fully remote)

1. Where they want to be in ~3 years (senior/staff AI PM at a large company / founder running

their own AI venture / director/head-of-AI track / deep technical specialist)

1. Timeframe they're working with (under 3 months / 3-6 months / 6-12 months / 12+ months,

still exploring)

Batch C — Action Plan, Roadmap, Risk:

1. Initiatives already underway or planned — multiSelect (certification/course / a side project

or startup / applying at scale / building a LinkedIn/content presence)

1. Resources they can put behind this (mostly their own time / savings or budget to invest /

collaborators or a small team / a mentor or community)

1. Biggest risk to the plan (runway/savings running out / burnout or losing motivation / market

conditions worsening / skills gap not closing fast enough)

1. Realistic weekly time commitment (under 5 hrs / 5-15 hrs / 15-30 hrs / 30+ hrs, treating it

like a second job)

Ask batch A first, then B, then C — don't front-load all twelve questions at once. Read each answer before framing the next batch's options (e.g. if they said "between roles," don't offer "employed full-time" as a challenge option later).

Step 3 — Ground the Market Analysis in reality, not vibes

The template's "Market and Competitive Analysis" section asks the user to research trends and competitor profiles manually. Don't leave that entirely on them — run a couple of WebSearch queries on current AI hiring trends for their target role/domain (e.g. "AI product manager hiring trends 2026", "[their domain] AI roles market demand") and fold real, dated findings into that section alongside whatever anecdotal insight they gave you. Cite what you find.

Step 4 — Write the Markdown strategy doc

Use the "Markdown output template" section near the bottom of this file as the structure — sequential numbering 1-7 (the source template the user supplied had a numbering typo skipping "3"; don't replicate that), and every heading and sub-heading from that template must appear in the output, in that order. Copy its section layout and replace every [bracketed] placeholder with this user's real content — don't leave a sub-bullet in as a stub if it doesn't apply, delete it instead.

This needs to be a genuinely detailed document, not a condensed summary — the point of the original template is that each section gets real thinking, the way the example template did ("Ai is must irrespective of time it takes | Money need to doubled | Location need to be CA" is specific and opinionated, not a placeholder-length one-liner). Write multiple sentences or several bullets per sub-section wherever the workshop answers support it — don't compress a rich answer into a single clipped line just to keep things short. Write in the user's voice where you can (pull phrasing from their actual answers rather than generic corporate language). Every section should reflect a real answer they gave, not a filler placeholder. Save as [name]-job-strategy.md in the outputs folder.

Step 5 — Build the HTML artifact

Design this one at runtime rather than filling in a fixed template — see "HTML artifact — design it at runtime" near the bottom of this file for the structural requirements and non-negotiables (self-contained single file, every number traceable to something the user actually said, etc.). The right layout depends on how much content this particular user has — someone with 8 KPIs and someone with 2 shouldn't get the same rigid grid.

Before doing anything else, deliver both files — call present_files with the .md and the .html together so the user has real, downloadable files in hand. This has to actually happen (not just be described) before you move on to Step 6; "the doc is done" means they can open it, not that you've described what it would contain.

Step 6 — Offer the follow-up menu (plain chat, not AskUserQuestion)

Once both files are delivered, offer the interview-prep follow-ups as a normal chat message — plain text, not the AskUserQuestion tool. This one's conversational by design: the user should be able to just reply "do #1 and #3" or "let's do the salary one" in their own words rather than clicking a multiple-choice widget. Something like:

Your strategy doc and visual artifact are ready above. Want to keep going with any of these?

1. Five stories for your first-round interview (I'll need the job description)

2. Project discussion key points

3. STAR story bank

4. Mock interview Q&A prep

5. Salary negotiation script

6. Company & interviewer research brief

Full instructions for each — what to ask for, what to research, and the output format — are in the "Follow-up menu — full details" section below. Read that section once the user picks one (or more) rather than improvising the flow from scratch, since each has a specific required input (e.g. #1 cannot proceed without a job description — don't skip that dependency). Within each follow-up flow, AskUserQuestion is still fine for the flow's own inputs (e.g. #2's "which project" question) — it's specifically the top-level "what do you want next" offer that should stay conversational.

Users may come back to this skill later just for a follow-up (no new strategy doc needed) — in that case skip Steps 1-5 and jump straight to the relevant flow below, using whatever resume/strategy-doc context already exists in the conversation.

Follow-up menu — full details

Read only the sub-section for what the user picked. Each one lists required inputs, what to ask if those inputs are missing, and the expected output. All outputs should be saved as Markdown files in the outputs folder ([name]-<followup>.md) — no need for an HTML artifact for these, the strategy doc is the one deliverable that gets the visual treatment.

1. Five Stories for First-Round Interview

Hard dependency: a job description. This flow cannot produce good output without one — don't generate generic stories and call it done. If the user hasn't already shared a JD in the conversation, stop and ask for it (paste the text, a link, or the job title + company so you can search for it) before doing anything else.

Once you have the JD:

1. Extract the 4-6 things it's clearly screening for (a mix of hard requirements and softer

signals — e.g. "cross-functional influence," "ambiguity," "technical depth," "shipped 0→1").

1. Cross-reference against the resume to find the strongest real story for each. If two

requirements point to the same story, that's fine — better a true, strong story reused than a thin invented one.

1. Write 5 stories in STAR format (Situation, Task, Action, Result), each explicitly tagged

with which requirement from the JD it's aimed at, 150-250 words each. Close each with a one-line "why this matters for this role" note.

Output: [name]-interview-stories.md, with the JD requirements listed at the top so the mapping is visible.

2. Project Discussion Key Points

Uses the resume as the primary reference. Before building this, ask (AskUserQuestion, one batch):

1. Which project(s) to focus on — offer the 2-3 most substantial ones you see on the resume as

options, plus "let me specify" via Other

1. What angle the interviewer is likely to probe — options: technical deep-dive (architecture,

trade-offs, how it actually works) / business impact (metrics, stakeholders, why it mattered) / leadership & process (how decisions got made, conflict, what they'd do differently) / not sure, cover all three

For each selected project, produce talking points covering: context (problem, constraints), your specific role and decisions, technical/product trade-offs you made and why, quantified outcome, and one honest "what I'd do differently" reflection — interviewers notice when candidates only tell the polished version.

Output: [name]-project-discussion-points.md.

3. STAR Story Bank

Broader and less JD-specific than #1 — a reusable bank rather than a targeted set. Pull from the full resume and build one story per competency, for the competencies that actually show up in the person's background (don't force a "conflict" story if nothing on the resume supports one — flag the gap instead so they know to prep it separately). Typical set: leadership, handling failure/setback, ambiguity, influence without authority, data-driven decision-making, cross-functional conflict.