Job Pipeline Os

Job Pipeline OS

A complete job search operating system that runs in local markdown files: pipeline tracking, tailored resumes and cover letters, interview prep, follow-up scripts, and visa-sponsorship-aware strategy. Uses a consistent data model per opportunity and supports add, tailor, prep, move, follow-up, weekly review, sponsorship check, and offer evaluation.

---

1. FIRST-TIME SETUP

If this skill has just been activated and no pipeline-config.md exists (or the user says "set up my pipeline"), ask the user once for the following and store the result in pipeline-config.md in the user's workspace or a designated folder:

| Field | Example | |-------|---------| | Full name | Jane Doe | | Target roles | Software Engineer, ML Engineer | | Target locations | San Francisco, Remote, New York | | Experience level | new grad / 1–3 yrs / 3–5 yrs / senior | | Visa status | US citizen / GC / OPT / H-1B / need sponsorship | | Target salary range | $X–$Y base (or "market") | | Deal-breakers | e.g., no startups under Series B; must be remote | | Resume text | User pastes full resume (stored for reuse) | | LinkedIn URL | profile link |

Confirm back and say: "Pipeline config saved. Use add [job URL] to add opportunities, tailor [company] for materials, prep [company] for interview prep, and weekly review for your dashboard."

---

2. DATA MODEL

Each opportunity is stored as opportunities/[company-slug].md. Use this exact structure (adapt section names if the user already has a different convention):

# [Company Name] — [Role Title]

## Details

- **URL:** [job posting URL]
- **Location:** [city / remote / hybrid]
- **Salary:** [range if listed]
- **Stage:** TARGET | APPLIED | PHONE_SCREEN | TECHNICAL | ONSITE | OFFER | ACCEPTED | REJECTED | WITHDRAWN
- **Applied:** YYYY-MM-DD
- **Last Updated:** YYYY-MM-DD
- **Visa Sponsorship:** YES | NO | UNKNOWN | ASKED-PENDING
- **Referral:** [name or NONE]
- **Source:** LinkedIn | company site | referral | recruiter

## Contacts

- [Name] — [Title] — [email or LinkedIn]

## Materials Used

- Resume version: v1 | v2 | custom
- Cover letter: yes | no
- Outreach: yes | no, date

## Timeline

- YYYY-MM-DD: [event, e.g., "Recruiter screen scheduled"]

## Notes

[Anything relevant]

## Interview Notes

[Questions asked, answers given, feedback received]

When creating or updating an opportunity file, use the company-slug (lowercase, hyphens, no spaces) as the filename, e.g. acme-corp.md, google.md.

---

3. COMMANDS

Implement each command as a workflow. Commands are case-insensitive; [company] can be company name or slug.

COMMAND: add [job URL or description]

1. Parse the job description (JD): Extract role title, company name, location, requirements (must-haves), nice-to-haves.
2. Detect sponsorship language:

• "must be authorized to work" / "no sponsorship" / "authorized to work in the US without sponsorship" → NO
• "will sponsor" / "visa sponsorship available" / "H-1B sponsorship" → YES
• No mention → UNKNOWN (tell user to investigate)

1. Create the opportunity file under opportunities/[company-slug].md with Stage = TARGET (or APPLIED if they already applied).
2. Output: "Added [Company] — [Role]. Sponsorship: [YES/NO/UNKNOWN]. Want me to generate tailored materials now?"

---

COMMAND: tailor [company]

1. Load the user's resume from pipeline-config.md (or resume-base.md / resume-ml.md / resume-backend.md as appropriate).
2. Load the opportunity file for [company].
3. Output:

• (a) Tailored resume bullets — Reorder and reword to match JD keywords; suggest which existing bullets to lead with.
• (b) Cover letter — Company-specific; reference their products, mission, or recent news; 3–4 short paragraphs.
• (c) Outreach message — For LinkedIn connection or referral request (1–2 short paragraphs).
• (d) Key skills to emphasize — List 5–7 from the JD to highlight in interview and materials.

Use [references/resume-templates.md](references/resume-templates.md) for ATS and formatting rules.

---

COMMAND: prep [company]

1. Load the opportunity file for [company].
2. Research the company (use web search): e.g., "[company] engineering culture values 2025", "[company] tech stack", "[company] recent news".
3. Output:

• (a) Company brief — Mission, product, recent news, tech stack.
• (b) 15 likely interview questions — Based on JD (behavioral + technical where applicable).
• (c) STAR stories — Map 3–5 stories from the user's resume to their requirements; suggest which story for which question.
• (d) 5 smart questions to ask the interviewer — Role-specific, not generic.
• (e) Red flags to watch for — Culture, stability, sponsorship vagueness.

---

COMMAND: move [company] [new stage]

1. Update the opportunity file: set Stage to the new stage (e.g., PHONE_SCREEN, TECHNICAL, ONSITE, OFFER, REJECTED, WITHDRAWN).
2. Update Last Updated to today.
3. If moved to REJECTED: Ask: "Want to send a post-rejection email to keep the relationship warm?" If yes, generate one using [references/outreach-scripts.md](references/outreach-scripts.md).
4. If moved to OFFER: Automatically run the offer evaluation workflow (see command evaluate offer [company]).

---

COMMAND: follow-up [company]

1. Load the opportunity file; check days since last contact (from Timeline or Applied date).
2. Generate the appropriate follow-up email based on stage:

• After application (e.g., 7 days): "Following up on my application for [role]…"
• After interview (24–48 hrs): Thank-you note with a specific callback to a discussion point.
• After final round (e.g., 5 days no response): Brief, professional status check.

1. Output the email and suggest adding a Timeline entry when they send it.

Use [references/outreach-scripts.md](references/outreach-scripts.md) for tone and templates.

---

COMMAND: weekly review

1. Scan all files in opportunities/ (all opportunity markdown files).
2. Aggregate by Stage: TARGET, APPLIED, PHONE_SCREEN, TECHNICAL, ONSITE, OFFER, ACCEPTED, REJECTED, WITHDRAWN.
3. Output in this format:

📊 JOB SEARCH WEEKLY REVIEW — [week of date]

PIPELINE HEALTH:
🎯 Targets: [n]
📤 Applied: [n]
📞 Phone screens: [n]
💻 Technicals: [n]
🏢 Onsites: [n]
💰 Offers: [n]

⚡ THIS WEEK'S ACTIONS:
- [Company]: follow up — 7 days since application
- [Company]: interview in 3 days — run prep command
- [Company]: offer expires [date] — evaluate now

🔴 STALE (no update in 14+ days):
- [list companies to follow up or remove]

💡 RECOMMENDATION:
[One line: e.g., "Apply rate is low — add 5 more targets" / "Interview-to-offer ratio is strong — focus on closing" / "2 offers expiring — prioritize evaluation"]

Optionally run scripts/pipeline-stats.py if the user has it, and include or summarize its output (application rate, response rate, interview rate, offer rate, bar chart).

---

COMMAND: sponsorship check [company]

1. Load [references/sponsorship-guide.md](references/sponsorship-guide.md).
2. Check: Is this company known to sponsor H-1B? (from guide or web: DOL LCA data, company careers page.)
3. Generate an email the user can send to HR/recruiter to ask about sponsorship without hurting candidacy:

Hi [name],

I'm excited about the [role] position and wanted to ask about visa sponsorship. I'm currently on [OPT/STEM OPT] and will need H-1B sponsorship in [timeframe]. Is this something [Company] supports for this role? I want to make sure we're aligned before we go further.

Thanks,
[Your name]

1. If company is unknown, suggest checking the USCIS LCA database or company careers page and offer to draft the email anyway.

---

COMMAND: evaluate offer [company]

1. Load [references/offer-evaluation.md](references/offer-evaluation.md).
2. Ask the user (or use data from the opportunity file if already there): base salary, bonus, equity (type, amount, vesting), benefits, location, remote policy, visa sponsorship plan (if applicable), start date.
3. Output:

• Compensation vs market: above / at / below (for role and level).
• Equity analysis: worth considering / not worth much at typical outcomes (brief caveat).
• Visa path: Will they sponsor? Timeline? Any risk?
• Total comp vs alternatives in pipeline (if any other offers or late-stage).
• Questions to ask before accepting (vesting, benefits, role scope, sponsorship in writing).
• Negotiation leverage points (e.g., competing offer, signing bonus, start date).

Do not invent market numbers; use ranges from the reference or say "compare to levels.fyi / Glassdoor for your level and location."

---

4. VISA SPONSORSHIP MODULE

Load [references/sponsorship-guide.md](references/sponsorship-guide.md) whenever the user's pipeline-config indicates they need sponsorship (OPT, H-1B, etc.).

Quick reference:

• Almost always sponsor: Google, Meta, Amazon, Microsoft, Apple, Netflix, Salesforce, Uber, Lyft, Airbnb, Stripe, most Big Tech.
• Usually sponsor (Series B+): Most well-funded startups with US entities.
• Rarely sponsor: Early-stage startups, many non-tech companies, agencies, some regional employers.

Timeline reality check for OPT holders:

• 12-month OPT: Must get H-1B (or other status) by end of OPT or leave/change status.
• STEM OPT (36 months total): Must get H-1B by end of 24-month extension if staying in US.
• H-1B lottery: Registration typically March; cap selection; filing by employer; start October 1 (fiscal year).
• If not selected: Company can re-register next year; consider cap-gap if on OPT when H-1B is filed.
• O-1: Alternative for "extraordinary ability"; some companies open to this for senior hires.

When adding or tailoring for a company, flag sponsorship (YES/NO/UNKNOWN) and suggest sponsorship check [company] or the HR email when UNKNOWN.

---

5. RESUME VERSION MANAGER

Track multiple resume versions so tailoring can start from the right base:

| File | Use case | |------|----------| | resume-base.md (or pipeline-config resume section) | Master / general version | | resume-ml.md | ML/AI-focused variant | | resume-backend.md | Backend/systems-focused variant | | resume-frontend.md | Frontend-focused variant (optional) |

When tailoring:

• If the JD is ML/AI-heavy → suggest starting from resume-ml.md.
• If backend/infra → resume-backend.md.
• Otherwise → resume-base.md or the main resume in pipeline-config.

In the opportunity file, record Resume version: v1 / v2 / custom so the user knows which variant was used for that application.

---

Scripts

scripts/pipeline-stats.py — When the user runs weekly review or asks for "pipeline stats", run this script if available. It reads all **opportunities/.md* files and outputs:

• Application rate, response rate, interview rate, offer rate
• ASCII bar chart of pipeline stages
• Average days per stage
• Benchmarks (e.g., "Many engineers apply to 100–150 companies, get ~10–15 screens, 3–5 onsites, 1–2 offers")

If the script is not present or fails, compute a simple count of stages from the opportunity files and show the weekly review format without the script output.

---

References

| File | When to use | |------|-------------| | [references/sponsorship-guide.md](references/sponsorship-guide.md) | Sponsorship check; visa timeline; how to ask HR; red flags | | [references/resume-templates.md](references/resume-templates.md) | Tailoring; ATS rules; bullet format (action + metric + impact) | | [references/outreach-scripts.md](references/outreach-scripts.md) | Cold/warm outreach; referral request; recruiter reply; post-rejection | | [references/offer-evaluation.md](references/offer-evaluation.md) | Evaluate offer; total comp; equity; negotiation; what not to say |

---

Notes for Skill Maintainers

• File locations: Assume opportunities live in opportunities/ and config in pipeline-config.md unless the user specifies otherwise. Create the folder/file when the user first runs add or setup.
• Company slug: Derive from company name (lowercase, replace spaces with hyphens, remove punctuation).
• Web search: Use for company research in prep and for sponsorship/LCA lookups when needed; do not invent sponsorship status.
• No live APIs: This skill works offline-first with markdown; no requirement to call external APIs except for optional web search for prep and sponsorship.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dikondaashish
• Source: dikondaashish/Claude-agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.