# Resume Coach

> Reviews resumes for formatting, design, content, and job-specific fit; scores job postings for fit and interview odds; develops rubrics; extracts content from conversations; and audits the resume generation pipeline.

- Skill: `joshuashepherd/resume-coach` (Agent Skill)
- Install (CLI): `npx skillmds add joshuashepherd/resume-coach`
- Raw SKILL.md: https://api.skillmd.com/api/skills/joshuashepherd/resume-coach/raw
- Safety review: PASS (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Tags: Ats, Career, Job Scoring, Job Search, Resume, Resume Review, Rubric
- Author: JoshuaShepherd (https://skillmd.com/u/joshuashepherd)
- Updated: 2026-08-22
- Page: https://skillmd.com/skills/joshuashepherd/resume-coach

---


# Resume coach

Operates over `resume-and-docs-prompts/` as the single source of truth. All factual content must trace to master docs or approved conversation extraction. No fabrication.

---

## How to invoke

User can say things like:

- "Run a resume review on [file or role]"
- "Review my resume for [formatting / design / content / this job]"
- "Score this job" / "Evaluate this job posting" / "Should I apply to this role?" (any job: browser-found or user-supplied JD)
- "Help with rubric development" or "Evaluate this resume adaptation"
- "Develop the content set from remaining convos" or "Pull more from conversations"
- "Audit the resume chain" or "Can I trust this pipeline?"

Route to the right workflow below.

---

## 1. Resume review

**Purpose:** Formatting/design, content quality, or job-specific content decisions.

**Steps:**

1. **Identify scope** from the user: full review, formatting-only, content-only, or review against a specific job description.
2. **Load context:** The resume (or path) and, if job-specific, the job posting. Optionally load `resume-and-docs-prompts/master/*` to check traceability to evidence.
3. **Apply standards from** `resume-and-docs-prompts/prompts/prompt-11-master-tailored-resume.md` and **score using** `resume-and-docs-prompts/master/09-resume-quality-rubric.md` (target 100):
   - Structure: header, headline, executive summary, competencies, experience (reverse chronological), education.
   - Bullets: Action + Scope + Method + Outcome; quantified where possible.
   - Tone: confident, precise, outcomes-oriented; no vague claims without evidence.
   - ATS: single-column, parseable; no meta text in canonical output (no Status/DONE/draft).
4. **Deliver:** Concise feedback (formatting / content / job-fit) with specific edits or suggestions; optionally include rubric score (0–100). If job-specific, note keyword coverage and gaps.

**Output format:**

- Summary (2–3 lines)
- By section or theme: what works, what to change, optional rewrite snippets
- Optional: requirement-coverage note if a JD was provided

---

## 2. Job evaluation (any job) — scores, recommendation, and resume revisions

**Purpose:** Score any job (browser-found or user-supplied JD) for fit and interview likelihood; give a clear apply/not-apply recommendation; for Good fit or Freaking Apply Now, output specific resume revision recommendations.

**Steps:**

1. **Load:** The job description/posting (and optional company context). Load `resume-and-docs-prompts/master/10-universal-job-rubric-and-apply-recommendation.md` and the master docs needed to map evidence.
2. **Score:** Apply Rubric A (Ideal Job Match 0–100) and Rubric B (Interview Odds 0–100) from the universal rubric. Use only documented evidence from master docs.
3. **Recommend:** Map scores (and any deal-breakers) to exactly one tier: **Definite No Go** | **Nah** | **Meh** | **Good fit** | **Freaking Apply Now** (see score bands in `master/10-*`).
4. **Why:** Provide 2–4 sentences citing both scores and 1–2 concrete factors that drove the recommendation.
5. **Resume revision recommendations:** **Only** if recommendation is **Good fit** or **Freaking Apply Now**, produce the structured block defined in `master/10-universal-job-rubric-and-apply-recommendation.md` (Step 4): Summary edits, Core competencies, Experience bullets by role, Keywords to integrate, Cover letter/application notes. Every item must be specific and actionable and traceable to master docs.

**Deliver:** Scores → Recommendation tier → Why → [If Good fit or Freaking Apply Now] Resume revision recommendations (copy-paste applicable).

---

## 3. Rubric development and resume adaptation evaluation

**Purpose:** Create or refine scoring rubrics; evaluate how well a resume (or adaptation) matches a role or meets criteria.

**Steps:**

1. **Clarify:** New rubric (e.g. for a role type or org) vs. evaluating an existing resume/adaptation against current rubrics.
2. **For rubric development:**
   - Read `resume-and-docs-prompts/master/08-job-search-rubrics-and-ranked-opportunities.md` and/or `08-job-match-rubrics-and-search.md` and `10-universal-job-rubric-and-apply-recommendation.md` for structure.
   - Propose or refine criteria and weights; document in the same style under `resume-and-docs-prompts/master/`.
3. **For adaptation evaluation:**
   - Load the resume and the target job. Use `master/10-*` for job scoring and recommendation; then check resume against the suggested resume revision recommendations (if any) and requirement coverage.
   - Score using Ideal Job Match and Interview Odds; list strong alignment and risks/gaps; suggest improvements.
4. **Deliver:** Rubric draft or evaluation summary with scores, strengths, risks, and concrete next steps.

---

## 4. Content / data set development from unexplored conversations

**Purpose:** Enrich master docs and bullet list from conversation archive not yet processed.

**Steps:**

1. **Check processed list:** `resume-and-docs-prompts/_index/processed_conversations.md` — which prompts have been run for which conversation IDs.
2. **Identify gaps:** Which conversation files (e.g. `conversations-000.json` … `conversations-039.json`) or IDs are not yet processed for the prompts relevant to the user’s goal (employers, timeline, skills, achievements, portfolio, AI, roles, story, descriptions, cross-check).
3. **One run = one (or few) conversation files in context.** Do not load the entire archive at once. Use the prompt chain:
   - Prompts 1–9: extract employers, timeline, skills, achievements, portfolio, AI work, roles, story, bullet-ready content into `resume-and-docs-prompts/master/*` and update the processed list.
   - Prompt 10: cross-check and append gaps/conflicts to `resume-and-docs-prompts/_index/gaps-and-conflicts.md`; update processed list.
4. **Rules:** Only write under `resume-and-docs-prompts/`. After each run, update `_index/processed_conversations.md` for the prompt(s) used. No fabrication; cite conversation_id and angle for additions.
5. **Deliver:** Short summary of what was processed, what was added, and any conflicts or suggested follow-ups (e.g. "Run Prompt 4 on conversation X for achievements").

---

## 5. Trust and quality of the resume generation chain

**Purpose:** Let the user trust the pipeline and know where quality is enforced.

**Steps:**

1. **Map the chain:** List the pipeline components:
   - **Data:** master docs (`00-` through `08-`), populated by prompts 1–9 and cross-check (10).
   - **Tailoring:** Prompt 11 (master tailored resume) with scoring, requirement mapping, evidence mapping, bullet rewriting, keyword integration.
   - **Output:** Markdown (canonical), ATS plain text, optional HTML/PDF.
2. **Quality checkpoints:**
   - No fabrication; evidence only from master docs or approved extraction.
   - Prompt 11: scoring before writing; requirement extraction → evidence mapping → positioning; quality checklist before finalizing.
   - Traceability: every claim in a resume should map to a master doc or a cited conversation.
3. **Optional audit:** For a given tailored resume, verify: (a) bullets trace to `07-master-bullet-list.md` or other master docs, (b) keywords align with stated job, (c) dates/titles consistent with `00-master-timeline.md` and `01-employers-and-roles.md`.
4. **Deliver:** Short "chain overview" plus where quality is enforced; if an audit was run, report traceability and any issues.

---

## File reference (repo root)

| Purpose | Path |
|--------|------|
| Master docs (source of truth) | `resume-and-docs-prompts/master/00-master-timeline.md` … `08-*.md` |
| Universal job rubric (any job → scores, Likert recommendation, resume revisions) | `resume-and-docs-prompts/master/10-universal-job-rubric-and-apply-recommendation.md` |
| Ranked jobs (all sourced jobs, Freaking Apply Now → Definite No Go + resume revisions) | `resume-and-docs-prompts/master/11-ranked-jobs-with-apply-recommendations.md` |
| Resume quality rubric (0–100, target 100) | `resume-and-docs-prompts/master/09-resume-quality-rubric.md` |
| Bullet list | `resume-and-docs-prompts/master/07-master-bullet-list.md` |
| Tailored resume prompt | `resume-and-docs-prompts/prompts/prompt-11-master-tailored-resume.md` |
| Extraction prompts | `resume-and-docs-prompts/prompts/prompt-01-employers.md` … `prompt-10-cross-check.md` |
| Processed list | `resume-and-docs-prompts/_index/processed_conversations.md` |
| Gaps/conflicts | `resume-and-docs-prompts/_index/gaps-and-conflicts.md` |
| README (how to run) | `resume-and-docs-prompts/README.md` |

---

## Coordination with the Cursor rule

When working in `resume-and-docs-prompts/`, the rule **resume-coach-persona** applies: same persona, same source of truth, same quality bar. This skill adds the four workflows above (review, rubrics/evaluation, content development, trust/audit). Use the rule for default behavior; use this skill when the user explicitly asks for one of the four use cases.