Menu
Debrief a voice simulation — analyze transcript against coached answers, track anti-patterns, log session
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Guided coaching conversation to discover your professional identity, strengths, values, and narrative patterns
Import an existing CV into structured data files — additive merge with existing data, can be run repeatedly
Multi-perspective deep-dive CV review (recruiter, hiring manager, competitor, skeptic, copy editor, source data auditor)
Fast quality-gate review of a generated CV against its target role
Scan a job portal for matching projects/roles, assess fit against profile, track evaluated ads
Generate a recruiter simulation prompt for the Claude App (voice mode) from a CV and job ad URL
| name | debrief |
| description | Debrief a voice simulation — analyze transcript against coached answers, track anti-patterns, log session |
| argument-hint | <path-to-cv> |
| user-invocable | true |
| allowed-tools | Read(*), Glob(*), Grep(*), Write(coaching/**), Edit(coaching/**) |
Analyze a recruiter screening transcript from a Claude App voice simulation. Compare the candidate's answers against coached answers, identify anti-patterns, rate performance, and log the session to the progress tracker.
The user has:
/voice-export)$ARGUMENTS (required): Path to the CV file used for the simulation (e.g. output/20260210-target-role-slug.md)The transcript is expected to be in the conversation context — the user pastes it before or after invoking the skill.
Load these files in parallel using the CV path to derive filenames:
*-cheatsheet.md matching CV filename (full file — this IS the coaching side, coached answers are needed here)coaching/coached-answers.md (general coached phrasings)*-DEEP-REVIEW.md matching CV filename (to know what tough questions were expected)coaching/progress-recruiter/_summary.mdcoaching/anti-pattern-tracker.md (global pattern status — which are resolved, which to watch for)framework/templates/recruiter-session.mdIf the transcript is not yet in the conversation, ask the user to paste it.
Break the transcript into Q&A pairs:
Note which deep review probing questions were actually asked by the recruiter, and which were skipped.
For each Q&A pair, assess:
Key insight: Real recruiters don't understand technical details, but they DO notice evasiveness, dodging, and red flags that signal "this person might embarrass me in the client interview."
For each answer, assess trust impact:
Special attention to:
Scan each answer for known anti-patterns. Load the full pattern list from coaching/anti-pattern-tracker.md § "Known Anti-Patterns Reference" — that file is the single source of truth for which patterns exist and their numbering. If the tracker has no patterns yet (new user), use these universal seed patterns:
Also watch for any NEW anti-patterns not yet tracked — add them to the tracker after the debrief.
Present the analysis to the user in this structure:
## Debrief — [Role Title] (Voice Simulation)
**Date:** [today]
**Questions asked:** [count]
**Overall confidence rating:** [1-5, based on answer quality average]
### Takeaway
[3-4 sentence executive summary: what happened in the simulation, what the dominant patterns were, what went well, and the single most important thing to fix next.]
### Recruiter Assessment Framework
Real recruiter screening has two dimensions:
| Dimension | Rating | What It Measures |
|-----------|--------|------------------|
| **Checkbox Match** | [1-5] | Did you hit the technical keywords from the job ad? |
| **Trust/Credibility** | [1-5] | "Will this candidate embarrass me if I send them to the client?" |
**Checkbox match:** [X/5] — [brief summary: e.g., "Hit all primary keywords, minor gap on [specific requirement]"]
**Trust/credibility:** [X/5] — [brief summary: e.g., "Strong except for dodging a concrete example on [topic]"]
**Likelihood of being forwarded:** [X/5] — Checkbox × Trust = overall outcome
**Likelihood of strong advocacy:** [X/5] — Would the recruiter champion you or just pass you along?
### Answer-by-Answer Analysis
| # | Topic | Question (summary) | Rating | Trust Impact | Anti-Patterns | Notes |
|---|-------|-------------------|--------|--------------|---------------|-------|
| 1 | Pitch | Self-introduction | 4/5 | ✅ | — | Strong opening, close to coached version |
| 2 | Technical | [Technical requirement] | 2/5 | ❌ | Didn't answer yes/no | Recruiter asked twice, creates doubt |
| ... | | | | | | |
### Anti-Patterns Triggered
- [x] Pattern name — specific example from transcript
- [ ] Pattern name — NOT triggered
### What Went Well
- [bullet points of strongest moments]
### What Needs Work
- [bullet points of weakest moments with coached alternatives]
### Strong Phrasings to Keep
- [any new strong phrasings worth saving to coached-answers.md]
### Deep Review Questions Coverage
- [which tough questions were asked, which weren't, how they were handled]
### Focus for Next Session
- [2-3 specific priorities based on this session's patterns]
Present the debrief report and discuss:
After the user confirms the assessment:
Create session file — copy framework/templates/recruiter-session.md to coaching/progress-recruiter/YYYY-MM-DD-HHMM-[role-slug].md. Include:
Update summary at coaching/progress-recruiter/_summary.md (if it doesn't exist yet, copy framework/templates/recruiter-summary.md first):
Update coached-answers.md if the user approved any new strong phrasings.
Update anti-pattern tracker at coaching/anti-pattern-tracker.md:
Data enrichment — check if the simulation surfaced new information (project details, achievements, technologies, skills) that should be captured in the data files. Follow the procedure in framework/data-enrichment.md.
If a session file for this date and role already exists (from a text-based coaching session), append a suffix:
2026-02-10-target-role-slug.md2026-02-10-target-role-slug-v2.md2026-02-11-target-role-slug-voice.md (prefer -voice suffix to distinguish from text sessions)