dcassil/resume-kit

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Show "what the ATS sees" for a resume — the detected sections, the extracted entities (name/contact/links, per-role dates → computed years-of-experience, degrees), and the zoned keyword breakdown, exactly as an ATS parser would read them. Read-only: it reports, it never edits the resume. Job-INDEPENDENT (no keyword…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Check a resume against generic, job-INDEPENDENT best practices — buzzwords, weak/duty openers, missing quantification, bullet quality — and report each finding classified autosuggestible vs needsuserinput. Read-only: it scores and reports, it never edits the resume. Drives the deterministic analyze-best-practices…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Check and produce ONLY the missing / injectable keyword list between a tailored resume, a proof surface, and a job description — which job keywords are missing from the tailored resume, which can be proved from a distinct master resume and/or confirmed Flow 1 learning-evidence (injectable), and which are not proved by…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Check resume↔job keyword coverage — the percentage of the job's keywords the resume already covers, plus the matched and missing keyword lists — for a resume against a specific job description. Presents ONLY the keyword slice; no composite / overall ATS score. Honors the project synonym index (aliasfile) so grown…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Check resume-only structural / parse issues that make an ATS mis-parse, miss, or error on a resume — section completeness plus deterministic structural recommendations (contact info, missing sections, dates, non-ASCII / formatting risks). Resume-ONLY: needs no job and produces NO keyword coverage and NO composite ATS…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Deterministically compare two resume versions (base vs candidate) against a job description and return a structured diff of their relative strengths. No LLM required.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Composite resume-intelligence workflow for producing a tailored, fitted, and exported resume from one prepared resume plus one or more jobs. Use when the user wants the complete end-to-end path or needs to understand the repeated use pattern across the four composable flows.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Render a ResumeDocument to a downloadable PDF or DOCX artifact. Fully deterministic and requires no LLM provider — always works with nollm. Uses ReportLab (PDF) and python-docx (DOCX) inside packages/export; no Chromium or upstream frontend involved.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Deterministically extract a list of CandidateEvidence records from a resume, optionally seeded with pre-approved claims. No LLM required.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Run the final job-aware fit and export over a tailored resume, enforce the rendered page hard gate, and do no preparation, job ingest, or tailoring. Best run in a subagent.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Parse a job and grow terminology learning against an existing prepared-resume learning base. Safe to run repeatedly for multiple jobs, and assumes Flow 1 has run. Best run in a subagent.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Interactive, truth-gated interview over a job's MISSING requirements. When tailoring finishes and the score is still below par (or a required must-have is uncovered), this skill walks the user through the missing items in the job's own priority order — required → preferred/nice-to-have → remaining unmatched keywords …

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Learn the user's outcome for an in-flight edit suggestion through the record-edit-feedback surface, then refresh preferences. Uses the structured EditFeedbackReasonCode enum plus an optional note, and records a PreferencePair when the outcome implies a preferred-vs-rejected comparison. No LLM required. Best run in a…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

The shared truth-gated growth loop for the project synonym index. When a deterministic ATS/match/gap run reports a job keyword as "missing" that the resume plausibly satisfies under a DIFFERENT surface term, this workflow proposes an alias, runs the truthfulness gate, asks the user to confirm, and ONLY THEN appends a…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Parse a job posting (pasted text, URL content, or a PDF/DOCX/MD/text file) into the canonical resume-kit JobDescription JSON, with structured requirements and skill keywords. This skill ORCHESTRATES a pipeline: for file inputs, resume-tool extract-text pulls the raw text deterministically (no LLM); pasted text / URL…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Parse a resume file (PDF, DOCX, Markdown, or plain text) into the canonical resume-kit ResumeDocument JSON, faithfully and losslessly. This skill ORCHESTRATES a deterministic pipeline: resume-tool extract-text pulls the raw text (no LLM), a confined interpretation subagent maps that text into the schema, and…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Final job-aware fit for a tailored resume: run the deterministic budget-fit capability, present ranked trim/compression candidates, drive decisions through the existing edit-session UX or use --auto-fit, and write the final resume only behind claim, ledger, and export page gates. No fabrication.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Prepare one resume without needing a job: produce an ATS-ready, no-custom canonical artifact for downstream tailoring, and seed durable full-resume learning from the source resume first so custom and unmapped content remain available as evidence. Best run in a subagent.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Rank truthful improvement candidates for a pending change with the deterministic rank-edit-candidates surface. The skill passes candidates, FeatureContext, optional preferences, and aliasfile to the CLI/MCP/facade surface, then presents the ranked candidates and reasons. It never edits the resume; chosen candidates…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Compatibility entry-point guide for the resume-intelligence end-to-end workflow. Use when a user asks for the older resume-workflow guide; it points to complete-resume-flow and the four composable flow skills rather than duplicating their stage gates.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Dispatch a subagent to review a resume against the original resume and the target job, then write STRUCTURED, parseable findings to resume-kit/review/ .md. The reviewer checks truthfulness (claims not supported by the original), missed JD requirements, over/under-claims, terminology-mirroring opportunities, and…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Seed and grow the project synonym index (aliasfile) automatically during the normal tailoring flow, so match / check-keywords / check-gaps / suggest-terminology are alias-aware from the very FIRST score instead of aliasfile being None on a fresh project. Runs after a resume + job are both active: proposes TRUTHFUL…

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Select the best-matching resume from a set of candidates for a given job description. No LLM required.

dcassil/resume-kit

Skill Claude CodeCodex

Part of resume-intelligence

Job-specific tailoring over a prepared resume, active job, and durable learning/evidence. Runs the truth-gated scoring, improvement, validation, and re-scoring loop after Flow 1 and Flow 2, then stops before perfect/export. Best run in a subagent.