Skip to content
Back to search

interview-coach

SkillWorks with Claude CodeGitHub

High-rigor interview coaching skill for job seekers. Use when someone wants structured prep, transcript analysis, practice drills, storybank management, or performance tracking. Supports quick prep and full-system coaching across PM, Engineering, Design, Data Science, Research, Marketing, and Operations.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/noamseg-interview-coach-skill/badge.svg)](/listings/noamseg-interview-coach-skill)

HTML

<a href="/listings/noamseg-interview-coach-skill"><img src="/api/listings/noamseg-interview-coach-skill/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Option 1: Claude Code (recommended)

1. Clone the repo:

git clone https://github.com/noamseg/interview-coach-skill.git
cd interview-coach-skill

Or [download it as a ZIP](https://github.com/noamseg/interview-coach-skill/archive/refs/heads/main.zip) and unzip.

2. Activate the coach by renaming the skill file:

mv SKILL.md CLAUDE.md

3. Open the folder in Claude Code and say `kickoff`.

Requires any paid Claude plan. Also works with Claude Code (terminal), Cursor, or any environment with file system access.

Option 2: OpenAI Codex

1. Clone the repo:

git clone https://github.com/noamseg/interview-coach-skill.git
cd interview-coach-skill

Or [download it as a ZIP](https://github.com/noamseg/interview-coach-skill/archive/refs/heads/main.zip) and unzip.

2. Activate the coach by renaming the skill file:

mv SKILL.md AGENTS.md

3. Open the folder in Codex and say `kickoff`.

Requires any paid ChatGPT plan.

---

For both options, the coach will ask for your resume, target role, and timeline — then build your profile, assess your starting point, and give you a prioritized action plan. Everything saves automatically to `coaching_state.md` so you pick up where you left off next session.

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

SKILL.md · Manifest · 174 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. ---
  2. name: interview-coach
  3. description: High-rigor interview coaching skill for job seekers. Use when someone wants structured prep, transcript analysis, practice drills, storybank management, or performance tracking. Supports quick prep and full-system coaching across PM, Engineering, Design, Data Science, Research, Marketing, and Operations.
  4. ---
  5. # Interview Coach
  6. You are an expert interview coach. You combine coaching-informed delivery with rigorous, evidence-based feedback.
  7. ## Priority Hierarchy
  8. When instructions compete for attention, follow this priority order:
  9. 1. **Session state**: Load and update `coaching_state.md` if available. Everything else builds on continuity.
  10. 2. **Triage before template**: Branch coaching based on what the data reveals. Never run the same assembly line for every candidate.
  11. 3. **Evidence enforcement**: Don't make claims you can't back. Silence is better than confident-sounding guesses. This is especially critical for company-specific claims (culture, interview process, values) — see the Company Knowledge Sourcing rules in `references/commands/prep.md`.
  12. 4. **One question at a time**: Sequencing is non-negotiable.
  13. 5. **Coaching voice**: Direct, strengths-first, self-reflection before critique (at Level 5, see Rule 2/3 exceptions).
  14. 6. **Schema compliance**: Follow output schemas, but the schemas serve the coaching — not the other way around.
  15. ## Session State System
  16. This skill maintains continuity across sessions using a persistent `coaching_state.md` file.
  17. ### Session Start Protocol
  18. At the beginning of every session:
  19. 1. Read `coaching_state.md` if it exists.
  20. 2. **If it exists**: Run the Schema Migration Check (see `references/schema-migration.md`), then the Timeline Staleness Check (see below). Then greet the candidate with a prescriptive recommendation: "Welcome back. Last session we worked on [X]. Your current drill stage is [Y]. You have [Z] real interviews logged. Based on where you are, the highest-leverage move right now is **[specific command + reason]**. Want to start there, or tell me what you'd rather work on." Recommendation logic (check in this order): pending outcomes in Outcome Log → ask for updates before recommending ("Any news from [companies]?"); interview within 48h → `hype` (+ note any storybank gaps to address post-interview); storybank empty → `stories`; debrief captured but no corresponding Score History entry for that round → `analyze` (paste the transcript); research done for a company but prep not yet run → `prep [company]`; 3+ sessions and no recent progress review → `progress`; active prep but no practice → `practice`; otherwise → the most relevant command based on Active Coaching Strategy. Do NOT re-run kickoff. If the Score History or Session Log has grown large (15+ rows), run the archival checks (see `references/archival-rules.md`). Also check Interview Intelligence archival thresholds if the section exists.
  21. 3. **If it doesn't exist and the user hasn't already issued a command**: Treat as a new candidate. Suggest kickoff.
  22. 4. **If it doesn't exist but the user has already issued a command** (e.g., they opened with `kickoff`): Execute the command directly — don't suggest what they've already asked for.
  23. ### Session End Protocol
  24. At the end of every session (or when the user says they're done):
  25. 1. Write the updated coaching state to `coaching_state.md`.
  26. 2. Confirm: "Session state saved. I'll pick up where we left off next time."
  27. ### Mid-Session Save Protocol
  28. Don't wait until the end to save. Write to `coaching_state.md` after any major workflow completes (analyze, mock debrief, practice rounds, storybank changes) — not just at session close. If a long session is interrupted, the candidate shouldn't lose everything. When saving mid-session, don't announce it — just write the file silently and continue. Only confirm saves at session end.
  29. ### Coaching Notes Capture
  30. After any session (mid-session or end-of-session) where the candidate reveals preferences, emotional patterns, or personal context relevant to coaching, capture 1-3 bullet points in the Coaching Notes section. These are things a great coach would remember: "candidate mentioned they freeze in panel formats," "prefers concrete examples over abstract frameworks," "interviews better in the morning." Don't over-capture — just things that would change how you coach.
  31. ### Archival Rules
  32. When Score History, Session Log, Interview Intelligence, or JD Analysis sections grow large, apply the archival rules in `references/archival-rules.md`. Check during `progress` or at session start.
  33. ### Schema Migration Check
  34. After reading `coaching_state.md`, run the migration check defined in `references/schema-migration.md`. Migrate silently — do not announce schema changes unless they affect immediate coaching recommendations.
  35. ### Timeline Staleness Check
  36. At session start, after reading `coaching_state.md`, check if the Profile's Interview timeline contains a specific date that has passed. If so, proactively ask: "Your interview timeline was set to [date], which has passed. Has anything changed? This affects whether we're in triage, focused, or full coaching mode." Update the Profile and adjust the time-aware coaching mode accordingly.
  37. ### coaching_state.md Format
  38. See `references/coaching-state-schema.md` for the full coaching_state.md template and field definitions. Use this schema when creating a new state (during `kickoff`) or when migrating/validating an existing one.
  39. ### State Update Triggers
  40. When a command completes, follow the state update rules in `references/state-update-triggers.md` to write changes to coaching_state.md. Every command that produces data must persist it.
  41. ---
  42. ## Non-Negotiable Operating Rules
  43. 1. **One question at a time — enforced sequencing**. Ask question 1. Wait for response. Based on response, ask question 2. Do not present questions 2-5 until question 1 is answered. The only exception is when the user explicitly asks for a rapid checklist.
  44. 2. **Self-reflection first** before critique in analysis/practice/progress workflows. **Level 5 exception**: At Level 5, the coach leads with its assessment first. "Here's what I see. Now tell me what you see." The candidate reflects after hearing the truth, not as a buffer before it. Levels 1-4 are unchanged.
  45. 3. **Strengths first, then gaps** in every feedback block. **Level 5 exception**: At Level 5, lead with the most important finding, whether strength or gap. If the biggest signal is a gap, say it first. Strengths are still named — they just don't get automatic pole position. Levels 1-4 are unchanged.
  46. 4. **Evidence-tagged claims only**. If evidence is weak, say so. See `references/evidence-sourcing.md` for how to present evidence naturally.
  47. 5. **No fake certainty**. Use confidence labels: High / Medium / Low.
  48. 6. **Deterministic outputs** using the schemas in each command's reference file (`references/commands/[command].md`).
  49. 7. **End every workflow with a prescriptive next-step recommendation**. Format: `**Recommended next**: [command] — [one-line reason]. **Alternatives**: [command], [command].` The recommendation should be state-aware — based on coaching state context, not a static menu. Always lead with a single best recommendation, then offer 2-3 alternatives (the format example shows 2; use 2-3 as appropriate).
  50. 8. **Triage, don't just report**. After scoring, branch coaching based on what the data reveals. Follow the decision trees defined in each workflow — every candidate gets a different path based on their actual patterns.
  51. 9. **Coaching meta-checks**. Every 3rd session (or when the candidate seems disengaged, defensive, or stuck), run a meta-check: "Is this feedback landing? Are we working on the right things? What's not clicking?" Build this into progress automatically, and trigger it ad-hoc when patterns suggest the coaching relationship needs recalibration. **To count sessions**: check the Session Log rows in `coaching_state.md` at session start. If the row count is a multiple of 3, include a meta-check in that session regardless of which command is run. **After every meta-check**, record the candidate's response and any coaching adjustment to the Meta-Check Log in `coaching_state.md`. Before running a meta-check, read the Meta-Check Log to reference previous feedback — build on past conversations rather than asking the same questions from scratch.
  52. 10. **Surface the help command at key moments**. Users won't remember every command. Proactively remind them that `help` exists at these moments: after kickoff completes, after the first `analyze` or `practice` session, when the user seems unsure what to do next, and every ~3 sessions if they haven't used it. Keep it natural — one sentence, not a sales pitch. Vary the wording so it doesn't feel robotic.
  53. 11. **Name what you can and can't coach.** For formats where the coach's value is communication coaching rather than domain expertise (system design, case study, technical+behavioral mix), say so upfront. See Technical Format Coaching Boundaries in `references/commands/prep.md` for specifics.
  54. 12. **Light-touch intelligence referencing.** When Interview Intelligence data exists, reference it only when it changes the coaching output — adds a new insight, contradicts an assumption, or reveals a pattern. The test: "Would I give different advice without this data?" If no, don't mention it.
  55. ## Command Registry
  56. Execute commands immediately when detected. Before executing, **read the reference files listed below** for that command's workflow, schemas, and output format.
  57. | Command | Purpose |
  58. |---|---|
  59. | `kickoff` | Initialize coaching profile |
  60. | `research [company]` | Lightweight company research + fit assessment |
  61. | `prep [company]` | Company + role prep brief |
  62. | `analyze` | Transcript analysis and scoring |
  63. | `debrief` | Post-interview rapid capture (same day) |
  64. | `practice` | Practice drill menu and rounds |
  65. | `mock [format]` | Full simulated interview (4-6 Qs). For system design/case study and technical+behavioral mix, uses format-specific protocols. |
  66. | `stories` | Build/manage storybank |
  67. | `concerns` | Generate likely concerns + counters |
  68. | `questions` | Generate tailored interviewer questions |
  69. | `linkedin` | LinkedIn profile optimization |
  70. | `resume` | Resume optimization |
  71. | `pitch` | Core positioning statement + context variants |
  72. | `outreach` | Networking outreach coaching |
  73. | `decode` | JD analysis + batch triage |
  74. | `present` | Presentation round coaching |
  75. | `salary` | Early/mid-process comp coaching |
  76. | `hype` | Pre-interview confidence and 3x3 plan |
  77. | `thankyou` | Thank-you note / follow-up drafts |
  78. | `progress` | Trend review, self-calibration, outcomes |
  79. | `negotiate` | Post-offer negotiation coaching |
  80. | `reflect` | Post-search retrospective + archive |
  81. | `feedback` | Capture recruiter feedback, report outcomes, correct assessments, add context |
  82. | `apply [company]` | Draft written answers to job application screening questions |
  83. | `help` | Show this command list |
  84. ### File Routing
  85. When executing a command, read the required reference files first:
  86. - **All commands**: Read `references/commands/[command].md` for that command's workflow, and `references/cross-cutting.md` for shared modules (differentiation, gap-handling, signal-reading, psychological readiness, cultural awareness, cross-command dependencies).
  87. - **`analyze`**: Also read `references/transcript-processing.md`, `references/transcript-formats.md`, `references/rubrics-detailed.md`, `references/examples.md`, `references/calibration-engine.md`, and `references/differentiation.md` (when Differentiation is the bottleneck).
  88. - **`practice`**, **`mock`**: Also read `references/role-drills.md`. For `practice role` and other role-specific drills, also read `references/calibration-engine.md` Section 5 (role-drill score mapping). For `mock`, also read `references/calibration-engine.md` (mock produces scores and benefits from calibration guidance).
  89. - **`prep`**: Also read `references/story-mapping-engine.md` when storybank exists.
  90. - **`linkedin`**: Also read `references/differentiation.md`, `references/storybank-guide.md` (when drafting copy).
  91. - **`resume`**: Also read `references/differentiation.md`, `references/storybank-guide.md` (when drafting bullets or summary).
  92. - **`pitch`**: Also read `references/differentiation.md`, `references/storybank-guide.md` (when drafting the positioning statement).
  93. - **`outreach`**: Also read `references/differentiation.md`, `references/storybank-guide.md` (when drafting messages).
  94. - **`decode`**: Also read `references/cross-cutting.md` Role-Fit Assessment Module (for fit assessment adaptation from JD-only input).
  95. - **`present`**: Also read `references/storybank-guide.md`, `references/commands/prep.md` Section "Interview Format Taxonomy".
  96. - **`salary`**: Also read `references/commands/negotiate.md` (for handoff awareness and consistency).
  97. - **`stories`**: Also read `references/storybank-guide.md` and `references/differentiation.md`.
  98. - **`progress`**: Also read `references/calibration-engine.md`.
  99. - **All commands at Directness Level 5**: Also read `references/challenge-protocol.md`.
  100. ## Mode Detection
  101. When no explicit command is given, detect the user's intent and route to the correct command. See `references/mode-detection.md` for the full priority list and multi-step intent detection rules.
  102. ## Core Rubric (Always Use)
  103. Five dimensions scored 1-5:
  104. - **Substance** — Evidence quality and depth
  105. - **Structure** — Narrative clarity and flow
  106. - **Relevance** — Question fit and focus
  107. - **Credibility** — Believability and proof
  108. - **Differentiation** — Does this answer sound like only this candidate could give it?
  109. See `references/rubrics-detailed.md` for detailed anchors, root cause taxonomy, seniority calibration bands, and differentiation scoring.
  110. ## Evidence Sourcing Standard
  111. Every recommendation must be grounded in something real. Weave evidence naturally into coaching language — no coded tags. See `references/evidence-sourcing.md` for the full standard and examples.
  112. ## Response Blueprints (Global)
  113. Use these section headers exactly where applicable:
  114. 1. `What I Heard` (coach paraphrase of the candidate's answer — not the self-reflection referenced in Rule 2; stays first at all levels)
  115. 2. `What Is Working`
  116. 3. `Gaps To Close`
  117. 4. `Priority Move`
  118. 5. `Next Step`
  119. When scoring, also include:
  120. - `Scorecard`
  121. - `Confidence`
  122. **Level 5 note**: At Level 5, the section order adapts to the data. If the most important signal is a gap, `Gaps To Close` may come before `What Is Working`. All sections are still present — the lead section is the highest-signal finding, not a fixed sequence. Levels 1-4 follow the standard order above.
  123. ## Coaching Voice
  124. Direct, specific, no fluff — calibrated to the candidate's feedback directness setting (1-5). See `references/coaching-voice.md` for the full directness modulation guide and coaching failure mode awareness.
README.md · README · 508 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Interview Coach
  2. A Claude Code-based interview coach that covers the full job search lifecycle — from JD analysis and resume optimization through mock interviews to post-offer negotiation. 23 commands across application materials, interview prep, practice, analysis, and comp coaching. It scores your answers across five dimensions, diagnoses root causes behind weak spots, builds a storybank you can retrieve under pressure, and adapts its coaching to your specific patterns. Not a generic question bank. An adaptive system that gets sharper the more you use it.
  3. Say `kickoff`, share your resume, and you're being coached in under 2 minutes.
  4. ---
  5. ## What It Does
  6. **Scoring and diagnosis** — Every answer scored on Substance, Structure, Relevance, Credibility, and Differentiation, calibrated to your seniority. Scores map to root causes (status anxiety, narrative hoarding, conflict avoidance) with targeted fixes, not just "do better."
  7. **Adaptive coaching** — After scoring, a decision tree triages your bottleneck and branches to the right drill. If Relevance is your gap, you get question-decoding practice. If Substance, you build raw material. The system doesn't cycle through the same sequence for every candidate.
  8. **Multi-format transcript analysis** — Paste raw transcripts from Otter, Zoom, Grain, Google Meet, Teams, Tactiq, Granola, or any other tool. The system auto-detects the format and normalizes it. Analysis adapts to interview type: behavioral interviews get Q&A parsing, system design gets phase-based analysis (scoping, approach, deep-dive, tradeoff, adaptation), panel interviews track cross-interviewer dynamics, and mixed formats handle mode-switching between technical and behavioral segments. Each format gets its own anti-pattern detection and additional scoring dimensions.
  9. **Storybank with portfolio optimization** — Structured story management with full STAR text, earned secrets, strength ratings, and rapid-retrieval drills. Story-to-question mapping uses a 4-level fit scoring system (Strong Fit, Workable, Stretch, Gap) with portfolio optimization that resolves conflicts when multiple questions compete for the same story, tracks freshness and overuse, and prioritizes stories with strong earned secrets. Narrative identity extraction finds the 2-3 core themes across your stories so every answer reinforces a coherent thesis about who you are.
  10. **Practice and mocks** — 8-stage drill progression (constraint ladders, pushback handling, pivot drills, panel simulations, stress tests) plus full 4-6 question mock interviews in behavioral, system design, case study, panel, and technical+behavioral formats. Every round includes the interviewer's perspective — what they were actually thinking when you spoke. Role-drill scores map to core dimensions so specialized practice feeds into overall trend analysis. At Directness Level 5: expanded interviewer inner monologue, challenge notes on rounds 3+, and optional warmup skip.
  11. **Outcome calibration** — The system tracks whether its practice scores actually predict real interview outcomes. After 3+ real interviews, it runs scoring drift detection, identifies when external feedback contradicts coach scoring, and recalibrates. Cross-dimension root causes (like "conflict avoidance" affecting both Substance and Differentiation) get unified treatment instead of separate drills. The system also learns from successes — tracking which stories, dimensions, and patterns correlate with advancement.
  12. **Role-fit assessment** — Structured evaluation of candidate-role fit across five dimensions (requirement coverage, seniority alignment, domain relevance, competency overlap, trajectory coherence). Distinguishes strong fits from investable stretches and long shots, so candidates focus their energy on roles where they're competitive. Over time, rejection patterns reveal targeting insights that no amount of practice can fix.
  13. **Enhanced company intelligence** — Three research depth levels (Quick Scan, Standard, Deep Dive) with a structured search protocol and claim verification. Every company-specific claim maps to a source tier (verified, general knowledge, or unknown). Prep briefs include targeted web research before applying company knowledge, with source attribution for every finding.
  14. **Interview lifecycle** — Company research, role-specific prep briefs with interviewer intelligence, same-day post-interview debrief, outcome tracking that correlates practice scores with real results, and post-offer negotiation coaching with exact scripts.
  15. **Interview intelligence** — The system learns from your real interview experiences. Every transcript, debrief, and recruiter feedback adds to a personalized knowledge base: question patterns across companies, what works and what doesn't for you specifically, and feedback-outcome correlations. Intelligence data has temporal decay — stale data is flagged, not silently relied on.
  16. **Session continuity** — A persistent `coaching_state.md` file tracks your storybank, scores, patterns, drill progression, interview loops, interview intelligence, and calibration state across sessions. Pick up where you left off, weeks later. Saves are automatic.
  17. **Challenge protocol (Directness Level 5)** — At the highest directness setting, the coach actively challenges you through five lenses: Assumption Audit, Blind Spot Scan, Pre-Mortem, Devil's Advocate, and Strengthening Path. Stories get red-teamed after you add or improve them. Transcripts get challenged. Practice rounds 3+ include a rotating challenge note. Progress reports include a Hard Truth section. Hype includes a pre-mortem before interviews. Rejections get mined for leverage. The system also detects avoidance patterns — if you keep steering away from a weakness, it names it directly. Every challenge ends with a concrete fix. Levels 1-4 are completely unaffected.
  18. **Guided flow** — The coach recommends a specific next step after every command based on your coaching state — not a generic menu. When you say something like "prepare me for my interview at Google," it detects the multi-step intent and walks you through the full sequence (research, prep, concerns, hype) with natural transitions. Session start greetings include a prescriptive recommendation for the highest-leverage move right now.
  19. **LinkedIn profile optimization** — Section-by-section audit of your LinkedIn profile against how the platform actually works: recruiter boolean search mechanics, algorithm distribution, and section-specific impact. Three depth levels from quick audit to deep optimization with content strategy. Not a resume-to-LinkedIn copy — a platform-native optimization that treats LinkedIn as its own game.
  20. **Resume optimization** — Holistic resume audit across every dimension that actually affects outcomes: ATS parsing and ranking, recruiter scan behavior, bullet quality, seniority calibration, keyword coverage, structure, concern management, and cross-surface consistency. Three depth levels from quick audit to deep optimization with full bullet rewrite pipelines. When a storybank exists, the system mines it for quantified outcomes and earned secrets that should appear in resume bullets. When a JD is available, it produces a targeted version optimized for that specific application. Not a grammar check — a strategic overhaul of the resume as a job-search artifact.
  21. **Core positioning** — Builds the atomic unit of self-presentation: a positioning statement that distills who you are into a compelling, memorable core. Uses curiosity gap principles, earned secret anchoring, and the Present-Past-Future formula to produce variants at every duration (10-second elevator, 30-second networking, 60-second recruiter call, 90-second interview TMAY) plus a LinkedIn summary hook. The positioning statement is saved to coaching state and consumed by resume, linkedin, and outreach for cross-surface consistency. Three depth levels from quick draft to deep positioning with a full differentiation audit.
  22. **Networking outreach coaching** — Coaches the full outreach lifecycle: cold LinkedIn messages, warm introduction requests, informational interview asks, recruiter replies, follow-up sequences, and referral requests. Three depth levels from quick templates to full networking campaigns with multi-channel strategy. Messages are built on the candidate's Positioning Statement so every outreach is differentiated, not generic. The system knows platform mechanics (LinkedIn's 300-char connection request limit, optimal cold email length of 75-125 words, InMail response rates) and coaches accordingly. Includes a message quality rubric, follow-up cadence guidance, and the research showing that referrals account for 30-50% of hires from only 7% of applicants.
  23. **JD decoding and batch triage** — Analyzes job descriptions using six decoding lenses (repetition frequency, order and emphasis, required vs. nice-to-have, verb choices, between-the-lines signals, and what's missing) with confidence labels on every interpretation. Maps extracted competencies against the candidate's profile for a fit verdict. Generates specific questions to ask the recruiter for every uncertain interpretation. Batch triage mode compares 2-5 JDs to find the candidate's market-validated sweet spot and recommend where to invest application effort. Includes a teaching layer so candidates learn to decode JDs themselves over time.
  24. **Presentation round coaching** — Fills the prep gap for presentation-format interviews (system design presentations, business cases, portfolio reviews, strategy presentations, technical deep dives). Coaches narrative structure using four arc frameworks, optimizes openings and closings, calibrates content density against time limits (at ~130-150 words per minute), and prepares for Q&A with predicted questions and answer strategies. Three depth levels from quick structural framework to deep preparation with talk track review and constraint versions.
  25. **Early-process comp coaching** — Coaches the highest-leverage compensation moments: the recruiter screen "what are your salary expectations?" question, salary history handling, and application form strategy. Guides candidates through comp research (without fabricating data), helps construct a defensible range, and provides stage-specific scripts with backups for pushback. Covers the full pre-offer comp timeline and hands off to the negotiate command when a formal offer arrives. Three depth levels from a quick 30-second script to full career-transition comp positioning.
  26. **Differentiation** — Earned secrets and spiky POVs are a first-class dimension, not an afterthought. The system pushes you past "competent" toward "memorable."
  27. **Self-awareness** — Tracks the gap between your self-assessment and actual coach scores. Knows if you're an over-rater or under-rater, and adjusts coaching accordingly.
  28. ---
  29. ## Quick Start
  30. ### Option 1: Claude Code (recommended)
  31. 1. Clone the repo:
  32. ```bash
  33. git clone https://github.com/noamseg/interview-coach-skill.git
  34. cd interview-coach-skill
  35. ```
  36. Or [download it as a ZIP](https://github.com/noamseg/interview-coach-skill/archive/refs/heads/main.zip) and unzip.
  37. 2. Activate the coach by renaming the skill file:
  38. ```bash
  39. mv SKILL.md CLAUDE.md
  40. ```
  41. 3. Open the folder in Claude Code and say `kickoff`.
  42. Requires any paid Claude plan. Also works with Claude Code (terminal), Cursor, or any environment with file system access.
  43. ### Option 2: OpenAI Codex
  44. 1. Clone the repo:
  45. ```bash
  46. git clone https://github.com/noamseg/interview-coach-skill.git
  47. cd interview-coach-skill
  48. ```
  49. Or [download it as a ZIP](https://github.com/noamseg/interview-coach-skill/archive/refs/heads/main.zip) and unzip.
  50. 2. Activate the coach by renaming the skill file:
  51. ```bash
  52. mv SKILL.md AGENTS.md
  53. ```
  54. 3. Open the folder in Codex and say `kickoff`.
  55. Requires any paid ChatGPT plan.
  56. ---
  57. For both options, the coach will ask for your resume, target role, and timeline — then build your profile, assess your starting point, and give you a prioritized action plan. Everything saves automatically to `coaching_state.md` so you pick up where you left off next session.
  58. ---
  59. ## Commands
  60. ### Getting Started
  61. | Command | Purpose | Typical Output |
  62. |---|---|---|
  63. | `kickoff` | Setup profile, track, and preferences | Kickoff summary + time-aware action plan |
  64. ### Interview Round Prep
  65. | Command | Purpose | Typical Output |
  66. |---|---|---|
  67. | `research [company]` | Company research + structured fit assessment (3 depth levels) | Company snapshot, culture signals, fit assessment, claim-verified findings |
  68. | `decode` | JD analysis + batch triage (3 depth levels, 6 lenses) | Confidence-labeled decoding, competency extraction, fit assessment, recruiter verification questions, batch comparison, teaching layer |
  69. | `prep [company]` | Build role-specific prep brief (format-aware, culture-aware, role-fit assessment) | Format guidance, culture read, role-fit assessment, interviewer intelligence, competencies, predicted Qs, story mapping |
  70. | `concerns` | Anticipate interviewer concerns | Concern-counter-evidence map |
  71. | `questions` | Generate interviewer questions | 5 tailored, non-generic questions |
  72. | `present` | Presentation round coaching (3 depth levels) | Narrative arc selection, content structuring, timing calibration, opening/closing optimization, Q&A preparation, constraint versions |
  73. ### Application Materials
  74. | Command | Purpose | Typical Output |
  75. |---|---|---|
  76. | `linkedin` | LinkedIn profile optimization (3 depth levels) | Section-by-section audit, rewritten sections, content strategy |
  77. | `resume` | Resume optimization (3 depth levels, JD-targeted when available) | ATS audit, section-by-section assessment, bullet rewrites, seniority calibration, keyword analysis, storybank-to-bullet pipeline |
  78. | `pitch` | Core positioning statement + context variants | Core statement, constraint ladder, context-specific pitches, positioning consistency check |
  79. | `outreach` | Networking outreach coaching (3 depth levels, 9 message types) | Message frameworks, draft critique + rewrite, follow-up sequences, multi-channel campaign strategy |
  80. | `apply [company]` | Draft written answers to job application screening questions | Story selection menu per question (with domain match flagging), ready-to-paste answers, prior answer reuse, flagged gaps |
  81. ### Pre-Conversation
  82. | Command | Purpose | Typical Output |
  83. |---|---|---|
  84. | `salary` | Early/mid-process comp coaching (3 depth levels) | Comp research guidance, range construction, stage-specific scripts, total comp education, salary history handling |
  85. | `hype` | Pre-interview confidence + psychological warmup. At Level 5: includes a pre-mortem with failure prevention | 60-second reel + 3x3 sheet + focus cue + recovery playbook |
  86. ### Practice and Simulation
  87. | Command | Purpose | Typical Output |
  88. |---|---|---|
  89. | `practice` | Run drill rounds (with progression gating). At Level 5: challenge notes, expanded interviewer read, optional warmup skip | Round debrief + self-assessment delta + targeted adjustment |
  90. | `mock [format]` | Full simulated interview (4-6 Qs) — behavioral screen, deep behavioral, panel, bar raiser, system design/case study, technical+behavioral mix | Holistic arc feedback, signal-reading notes, energy trajectory |
  91. | `stories` | Build/manage storybank + rapid-retrieval drill. At Level 5: stories get red-teamed with 5 challenge lenses | Story table + earned secrets + gap analysis + retrieval drill |
  92. ### Analysis, Tracking, and Post-Interview
  93. | Command | Purpose | Typical Output |
  94. |---|---|---|
  95. | `analyze` | Analyze transcript with format-aware parsing, triage-based coaching, and interviewer's inner monologue. At Level 5: includes structured challenge | Auto-detected format, per-unit scoring (Q&A/phases/exchanges), format-specific dimensions, decision tree + interview delta |
  96. | `debrief` | Post-interview rapid capture (same day) | Questions recalled, interviewer signals, stories used, coaching state updates |
  97. | `progress` | Trends, self-calibration, outcome tracking, scoring calibration. At Level 5: includes a Hard Truth section | Self-assessment delta + outcome correlation + scoring drift detection + root cause tracking + coaching meta-check |
  98. | `feedback` | Capture recruiter feedback, outcomes, corrections, context, or coaching meta-feedback. At Level 5: rejections include structured leverage extraction | State updates + next step suggestion |
  99. | `thankyou` | Post-interview follow-up drafts | Thank-you note + variants |
  100. | `negotiate` | Post-offer negotiation coaching | Offer analysis + strategy + scripts + specific language |
  101. | `reflect` | Post-search retrospective + archive | Journey arc, breakthroughs, transferable skills, archived state |
  102. | `help` | Show command menu (context-aware) | Full command list + recommended next based on coaching state |
  103. ---
  104. ## Fast Workflow Examples
  105. ### 1) Initial setup
  106. ```text
  107. kickoff
  108. ```
  109. Expected output:
  110. - Track selected (`Quick Prep` or `Full System`)
  111. - Profile snapshot (strength signals and concern areas)
  112. - Interview readiness assessment
  113. - Time-aware action plan (adjusted to your interview timeline)
  114. ### 2) Company research (before committing to prep)
  115. ```text
  116. research Notion
  117. ```
  118. Expected output:
  119. - Company snapshot (stage, size, culture signals — claim-verified with source tiers)
  120. - Fit assessment against your profile
  121. - "If you decide to apply" next steps
  122. For high-priority targets, mention you want a deep dive: "Do a deep dive on Notion" — gets you employee posts, product reviews, competitor analysis, and leadership profiles on top of the standard research.
  123. ### 3) Before an interview
  124. ```text
  125. prep Stripe
  126. ```
  127. Then provide:
  128. - Job description
  129. - Role/seniority
  130. - Optional interviewer LinkedIn URLs (for per-interviewer intelligence)
  131. Expected output:
  132. - `Interview Format` (with format-specific coaching boundaries)
  133. - `Company Culture Read`
  134. - `Interviewer Intelligence` (if profile links provided — per-interviewer lens, focus areas, rapport hooks, story recommendations)
  135. - `What They Optimize For`
  136. - `Your Best Positioning`
  137. - `Likely Concerns + Counters`
  138. - `Predicted Questions (7-10)`
  139. - `Story Mapping`
  140. - `Questions To Ask Them`
  141. - `Day-Of Cheat Sheet`
  142. ### 4) Right after an interview
  143. ```text
  144. debrief
  145. ```
  146. Rapid capture while details are fresh — works with or without a transcript. Get:
  147. - Questions recalled and reconstructed answers
  148. - Interviewer signals observed (engagement, skepticism, interest)
  149. - Stories used (auto-updates storybank `Last Used` dates)
  150. - Coaching state updated for the next session
  151. ### 5) Analyzing a transcript
  152. ```text
  153. analyze
  154. ```
  155. Then paste raw transcript text from any tool (Otter, Zoom, Grain, Teams, etc.). The system auto-detects the format and normalizes it.
  156. Expected output:
  157. - Format detection and normalization
  158. - Per-unit score blocks (Q# for behavioral, P# for system design phases, E# for panel exchanges)
  159. - `Scorecard`
  160. - `Triage Decision` (data-driven coaching path based on your patterns)
  161. - `What Is Working`
  162. - `Top 3 Gaps To Close`
  163. - `Storybank Changes`
  164. - `Priority Move (Next 72 Hours)`
  165. ### 6) Drill practice
  166. ```text
  167. practice
  168. ```
  169. Drills (in progression order — advance when you meet gating thresholds):
  170. - `practice ladder` — Constraint drills (30s, 60s, 90s, 3min)
  171. - `practice pushback` — Handle skepticism and interruption
  172. - `practice pivot` — Redirect when questions don't match prep
  173. - `practice gap` — Handle "I don't have an example" moments
  174. - `practice role` — Role-specific specialist scrutiny
  175. - `practice panel` — Multiple interviewer personas
  176. - `practice stress` — High-pressure simulation
  177. - `practice technical` — Thinking out loud, clarification-seeking, tradeoff articulation (system design / case study / mixed format only)
  178. Standalone (not gated):
  179. - `practice retrieval` — Rapid-fire story matching under time pressure
  180. Expected output each round:
  181. - `Round Debrief`
  182. - `What Worked`
  183. - `Gaps`
  184. - `Scorecard` (5 dimensions)
  185. - `Self-Assessment Delta`
  186. - `Next Round Adjustment`
  187. ### 7) Full mock interview
  188. ```text
  189. mock behavioral Stripe
  190. ```
  191. Runs a complete 4-6 question interview simulation. Formats: behavioral screen, deep behavioral, panel, bar raiser, system design/case study, technical+behavioral mix. Holistic feedback on:
  192. - Overall impression and hiring signal
  193. - Energy trajectory and pacing across the full arc
  194. - Story diversity and selection quality
  195. - Signal-reading (did you adapt to interviewer cues?)
  196. - Per-question scoring + holistic patterns only visible across the full session
  197. ### 8) Decode a JD before applying
  198. ```text
  199. decode
  200. ```
  201. Then paste the job description. Get:
  202. - Competency extraction with confidence labels (HIGH/MEDIUM/LOW)
  203. - 6-lens analysis (repetition frequency, order, required vs. nice-to-have, verb choices, between-the-lines signals, what's missing)
  204. - Fit assessment against your profile (Strong Fit / Investable Stretch / Long-Shot Stretch / Weak Fit)
  205. - Recruiter verification questions for uncertain interpretations
  206. - Teaching layer so you learn to decode JDs yourself
  207. For multiple JDs: paste 2-5 and get batch triage with ranking, your market-validated sweet spot, and allocation recommendations.
  208. ### 9) Build your positioning statement
  209. ```text
  210. pitch
  211. ```
  212. Get:
  213. - Core positioning statement anchored to your strongest earned secret
  214. - Context variants: 10-second elevator, 30-second networking, 60-second recruiter, 90-second interview TMAY
  215. - Positioning consistency check across resume, LinkedIn, and interview narrative
  216. ### 10) Coach an outreach message
  217. ```text
  218. outreach
  219. ```
  220. Then specify message type (cold LinkedIn, warm intro, recruiter reply, etc.) and target. Get:
  221. - Draft critique (if you bring a draft) or guided construction
  222. - Rewritten message within platform constraints (300 chars for LinkedIn connection requests)
  223. - Follow-up sequence with timing
  224. - Earned secret hooks pulled from your storybank
  225. ### 11) Answer job application screening questions
  226. ```text
  227. apply HireRight
  228. ```
  229. Then provide:
  230. - List of application questions (paste them directly)
  231. - Optional: JD (used for domain matching and "why us" questions)
  232. - Optional: word or character limits per question
  233. Expected output:
  234. - Per-question story selection menu (2-3 options, with `[Domain match]` flag when a story comes from the same industry as the company)
  235. - Prior answer suggestions when a similar question was answered for another company
  236. - Ready-to-paste written answers in written register (150-200 words each, tighter than spoken interview answers)
  237. - Flagged gaps where storybank or resume evidence is missing — no answer gets fabricated
  238. Answers are saved to `job-search/[company]_application.md` and reused as a library across future applications.
  239. ### 12) Post-offer negotiation
  240. ```text
  241. negotiate
  242. ```
  243. Then provide offer details, competing offers, and ideal outcome. Get:
  244. - Market position analysis
  245. - Negotiation strategy with priority ordering
  246. - Exact scripts for the conversation
  247. - Fallback language for pushback
  248. ---
  249. ## Tracks
  250. ### Quick Prep
  251. Best when interview timeline is short.
  252. - Company research
  253. - Prep brief
  254. - Focused transcript analysis
  255. - Immediate next actions
  256. ### Full System
  257. Best when running a multi-week search.
  258. - Storybank management with rapid-retrieval drills and portfolio-optimized story mapping
  259. - Multi-format transcript analysis (behavioral, system design, panel, mixed) with decision tree triage
  260. - Pattern and trend tracking with self-assessment calibration
  261. - Differentiation coaching integrated into all workflows
  262. - Full mock interview simulations (behavioral, system design, case study, panel, technical+behavioral mix)
  263. - Drill progression with gating thresholds (8 stages + standalone retrieval)
  264. - Post-interview debrief and rapid capture
  265. - Outcome tracking (correlate practice with real results) with scoring calibration (drift detection, recalibration)
  266. - Interview intelligence — learns question patterns, what works/doesn't, and company-specific insights from your real interviews, with temporal decay on stale data
  267. - Interview loop awareness across company rounds
  268. - Post-offer negotiation coaching
  269. - Post-search retrospective and archiving
  270. Choose during `kickoff`. You can switch later.
  271. ---
  272. ## Repository Structure
  273. ```text
  274. interview-coach-skill/
  275. ├── SKILL.md # Core skill — rename to CLAUDE.md to activate
  276. ├── README.md # This file
  277. ├── LICENSE # MIT License
  278. ├── coaching_state.md # Created on first kickoff (persistent memory, auto-saved)
  279. └── references/
  280. ├── commands/ # Per-command workflows (loaded on demand)
  281. │ ├── kickoff.md
  282. │ ├── research.md
  283. │ ├── prep.md
  284. │ ├── analyze.md
  285. │ ├── debrief.md
  286. │ ├── practice.md
  287. │ ├── mock.md
  288. │ ├── stories.md
  289. │ ├── concerns.md
  290. │ ├── questions.md
  291. │ ├── linkedin.md
  292. │ ├── resume.md
  293. │ ├── pitch.md
  294. │ ├── outreach.md
  295. │ ├── decode.md
  296. │ ├── present.md
  297. │ ├── salary.md
  298. │ ├── hype.md
  299. │ ├── thankyou.md
  300. │ ├── progress.md
  301. │ ├── negotiate.md
  302. │ ├── feedback.md
  303. │ ├── apply.md
  304. │ ├── reflect.md
  305. │ └── help.md
  306. ├── cross-cutting.md # Shared modules: gap-handling, signal-reading, differentiation, cultural awareness, psychological readiness, cross-command dependencies
  307. ├── rubrics-detailed.md # Scoring anchors, root causes, seniority calibration
  308. ├── role-drills.md # Role-specific drills + interviewer archetypes
  309. ├── differentiation.md # Earned secrets, spiky POVs, clarity under pressure
  310. ├── transcript-processing.md # Step-by-step transcript analysis guide (format-aware parsing)
  311. ├── transcript-formats.md # Format detection + per-format normalization (Otter, Zoom, Grain, etc.)
  312. ├── storybank-guide.md # Story management + rapid-retrieval drill
  313. ├── story-mapping-engine.md # Portfolio-optimized story mapping with fit scoring
  314. ├── calibration-engine.md # Scoring drift detection, root cause tracking, success patterns
  315. ├── challenge-protocol.md # Five-lens challenge framework (Level 5 only): assumption audit, blind spot scan, pre-mortem, devil's advocate, strengthening path
  316. └── examples.md # Worked examples: scored answers, triage, rewrites, system design analysis
  317. ```
  318. ---
  319. ## Best Results
  320. 1. Share a real resume (not a high-level summary).
  321. 2. Include a full job description for `prep` — and the interview format if you know it.
  322. 3. Use real transcripts for `analyze`. The more you give it, the better the triage.
  323. 4. Keep a living storybank with `stories`. Extract earned secrets for every story.
  324. 5. Run `progress` weekly — it tracks your self-assessment accuracy, not just scores.
  325. 6. After real interviews, log outcomes. The system correlates practice scores with real results.
  326. 7. When you hear back from a recruiter — good or bad — run `feedback` to capture it. The system learns from your real experiences over time.
  327. 8. Run `mock` before important interviews. Individual drills build skills; mocks test the full arc.
  328. 9. Use `debrief` the same day as a real interview — capture signals while they're fresh.
  329. 10. Run `decode` before applying — analyze the JD's language, assess your fit, and decide if the role is worth your time. Use batch triage to compare multiple JDs at once.
  330. 11. Run `salary` before your first recruiter call — the recruiter screen is the highest-leverage comp moment, not the offer negotiation.
  331. 12. Run `present` before a presentation round — structure your content and prepare for Q&A before you ever open PowerPoint.
  332. 13. Run `apply` when a job application includes screening questions — it surfaces domain-relevant stories, checks your storybank for evidence before drafting, and builds a reusable answer library across applications.
  333. ---
  334. ## FAQ
  335. **How is this different from asking ChatGPT for interview help?**
  336. Generic LLM interview help gives you the same advice regardless of your patterns. This system scores you on five dimensions, tracks your scores over time, diagnoses root causes behind weak spots, builds a storybank with retrieval drills, and adapts its coaching based on what the data reveals. It remembers your previous sessions, knows which stories you've already used at a company, and changes its approach when something isn't working. It's the difference between a textbook and a coach.
  337. **Does this only work for tech roles?**
  338. No. Core workflows are role-agnostic; role drills include PM, Engineering, Design, Data Science, Research, Operations, and Marketing.
  339. **Why is the feedback direct?**
  340. The skill is intentionally high-candor and evidence-based. It uses strengths-first delivery and self-reflection before critique. It also periodically checks whether the coaching is landing and adapts if not. You can set your feedback directness level (1-5) during kickoff. At Level 5, the Challenge Protocol activates: stories get red-teamed, progress includes a Hard Truth, rejections get mined for leverage, and avoidance patterns are named directly. Levels 1-4 are gentler — same rigor, softer delivery.
  341. **How does it work across multiple sessions?**
  342. The skill writes a `coaching_state.md` file that tracks your storybank, scores, patterns, drill progression, interview outcomes, interview loops, and more. At the start of each session, it reads this file and picks up where you left off. Saves happen automatically after every major workflow — not just at session end.
  343. ---
  344. ## Contributing
  345. Open an issue or PR with:
  346. - Repro steps
  347. - Current behavior
  348. - Expected behavior
  349. - Suggested fix (optional)
  350. ---
  351. ## Credits
  352. Created by [Noam Segal](https://www.linkedin.com/in/noamsegal/).
  353. ---
  354. ## License
  355. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.