jobclaw-skills

Build or update the JobClaw master profile — the single file every other job-hunt skill reads. Ingests a resume (extracting every embedded URL), LinkedIn PDF, exported AI-memory, and links, then asks only for what's missing in one batched, optional gap prompt. Captures career-goal context for downstream skills.

Orchestrate the full apply chain for one job — capture the job once, score fit, tailor the resume (PDF), write the cover letter, and answer the application questions — in auto-pilot (run end-to-end, pause only at decision/send gates) or review (confirm each step) mode. Stops before anything that sends or submits. The single entry point for "apply to this job".

Render a resume to a polished PDF via rendercv (Typst engine). Exposes all 9 built-in themes (engineeringresumes is the ATS-safe default) and embeds every link as a real, clickable PDF annotation. Reads the tailor-resume sidecar so the tailored content and links survive intact.

Produce a job-specific, ATS-optimized resume from the user's master profile and a target job — reordering, reframing, and surfacing the most relevant real experience and keywords. Enforces a content rubric (bullet shape, length, parallel grammar, metric coverage), then renders a polished PDF by default (via render-resume) and visually QAs it. Never fabricates.

Answer one or many application-form questions from the master profile, obeying per-type format rules — numeric → number only, yes/no → Yes|No, short → ≤ 2 sentences, paragraph → ≤ max_chars, enum → one of options. Batch JSON in/out (for JobClaw form-fill); single-question paste mode for interactive use. Flags any question without profile evidence rather than fabricating an answer.

Coach-style multi-turn conversation grounded in the master profile. The ONE skill allowed to read AND paraphrase (never quote) the profile's `context:` block — career_goal and additional_info — to direct advice. Last 8 messages of history retained.

Coach a counter-offer — leverage analysis, floor / target / walk-away ranges per comp component, and a draft counter email the human sends (never auto-send). Reads companies/<slug>.json for market comp when present; otherwise asks for an inline range or runs a thin WebSearch. Region pack drives currency / structure (USD base+equity vs INR LPA CTC fixed+variable).

Draft a reply email matched to one of the taxonomy's reply intents — reply-schedule, reply-accept (with sub-templates for scheduling / assessment / offer), reply-decline, reply-ask, reply-info, follow-up. Mirrors inbound tone; interpolates dates / links; uses the profile's signature. Never auto-sends.

Find relevant job listings for the user by combining multiple sources — Claude web search, ATS-direct boards (Greenhouse/Lever/Ashby/Workday/SmartRecruiters), Hacker News hiring, and optional APIs (Adzuna, SerpApi) with a Firecrawl scrape fallback — then normalize, dedupe across runs, and rank against the master profile.

Resolve the canonical application status from signals (triage class + current status + days-since-apply / days-since-last-contact + optional portal state) using the taxonomy decision table. Emits {status, next_action, next_action_due_on, rationale}; optionally updates a flat tracker.json. Honours the offer-accept guardrail and the ghosting rule.

Map a transition from a current role to a target role — surface real LinkedIn-style transition examples + a personalized gap roadmap (skills, experiences, signals) grounded in the user's profile. Optional Google CSE script (better recall) when keys are present; WebSearch fallback ships keyless.

Run a text-based mock interview — one question at a time, capture the user's verbatim answer, end with a scored report per the shared report-schema. Defaults to 5 questions (matches Go original); configurable up to 10. Unanswered questions score 0. Forked.

Build an interview prep brief — 8–12 likely questions split across the loop's rounds, profile-grounded STAR talking points per question, a prep checklist, and diplomatic questions to ask back. Reads companies/<slug>.json from research-company if present; otherwise runs a thin in-line sweep. Forked context.

Decide whether a blocking situation needs human input; compose a mobile-first prompt with ≤ 3 quick-reply options + a stated timeout default; parse a free-text reply back into a structured decision. v0 reasoning-only — delivery (Telegram / WhatsApp) lives in JobClaw's HITL MCP, not here. Always persists requests/<id>.json in compose mode.

Build a structured company × role brief — compensation, day-to-day, career growth, recent news, community sentiment (Reddit / Blind / Glassdoor), red flags, and a "fit for this user" view. Heavy reasoning — runs forked. Writes companies/<slug>.{md,json} sidecars consumed by prep-interview and coach-negotiation.

Visually QA a rendered resume PDF and auto-fix layout/content defects in a bounded loop: render → screenshot → deterministic gate → vision rubric → apply one fix → re-render, until it passes or hits a pass cap. Catches widows, under-fill, overflow, orphaned headers, and imbalance that text-only logic can't see.

Score how well the user's master profile fits a specific job — an ATS-style match percentage with matched/missing keywords, strengths, gaps, and a clear apply-or-skip recommendation. Optionally writes a structured sidecar that tailor-resume consumes.

Classify a recruiting email (or a batch) into one of the 10 canonical email classes, extract structured fields (company / role / dates / links / asks / deadline / sentiment), and suggest the next action. Writes inbox/triage-<YYYY-MM-DD>.json — consumed by infer-status and draft-reply.

Produce a job-specific, profile-grounded cover letter in markdown — ≤ 350 words, ≤ 1 page, three paragraphs (hook → two quantified proofs → fit close). Reuses keywords from a prior tailor-resume sidecar when present. Writes a JSON sidecar reserved for a future /render-resume --kind cover-letter (rendercv).

Write a short, profile-grounded cold outreach message to a recruiter, hiring manager, or potential referrer — LinkedIn DM or email — with one clear CTA and no flattery.