← ClaudeAtlas

apply-to-joblisted

Run the entire application pipeline for a specific job from one command — extract the JD, assess fit, tailor the LaTeX resume, render the PDF, draft any free-text answers the form asks for, and prepare the submission for your confirmation. Use whenever the user gives a job URL or JD and wants the end-to-end result rather than a single step. If the user only wants to analyze a JD, check fit, or tailor without rendering, prefer the narrower skills (extract-jd / resume-fit-report / tailor-resume) instead. Orchestrates those skills in order, pauses for review before rendering, and never submits without an explicit yes.
DhruvPrajapati4/apply-kit · ★ 0 · Data & Documents · score 70
Install: claude install-skill DhruvPrajapati4/apply-kit
# apply-to-job One entry point that runs the full pipeline for a job posting. Invoke the niche skills in order; each writes its artifact to the session scratchpad. ## Usage `/apply-to-job <job-url | pasted JD | path-to-jd-file>` ## Delegation (scoped subagents) Each stage has a matching subagent with a deliberately narrow tool allowlist, so a stage cannot overstep even if pushed (defense-in-depth behind the guardrails). Prefer dispatching each step to its subagent: - discovery (optional) → `job-scout` (searches ATS boards, cannot edit the resume) - step 0 → `resume-ingestor` (reads a resume file, no web) - step 1 → `jd-extractor` (web access, no editing/shell) - step 2 → `fit-analyst` (read-only, no web) - step 3 → `resume-tailor` (edit scratchpad only, no web, no shell) - step 4 → `resume-renderer` (compile only) - step 5 → `question-answerer` (drafts answers, no web, no shell) - step 6 → no subagent, by design. Submission is irreversible and stays in the main conversation where the user can see and stop it. Running the skills inline is also fine; the subagents exist to enforce tool scope. ## Pipeline This skill handles one known posting. If the user has no posting yet, prefer `find-and-apply`, which searches, scores and shortlists first and then runs this pipeline over each job they pick. Running `find-jobs` and feeding a lead in here by hand works too. 0. **ingest-resume** (when the user provides a resume file, or no working master is set yet) → `master.tex` + `