Executive Application Helper

A complete, battle-tested framework for running senior-executive job applications and interview preparation as a Claude project — with an evidence bank, matrix-based retrieval, verification skills that gate quality mechanically, and data-integrity disciplines earned the hard way.

Built and iterated across months of real executive applications (CIO / CTO / COO / CDO / SES and equivalent roles). Every rule in this framework exists because something went wrong without it.

What this is — and what it deliberately is not

This is not an AI application generator. It will never apply for you, and it is not designed to.

The point of this framework is not to mass-produce applications at the push of a button — that path produces generic, AI-slop submissions that say nothing true about you and that any experienced panel can spot in seconds. This framework is the opposite of that. Its job is to help you build a rigorous, structured bank of your own real evidence — your actual proof points, metrics, scope, and outcomes — and then assemble the strongest possible application from your own material, tailored to a specific role.

Three principles are non-negotiable and built into how it works:

1. Your evidence, not invented evidence. Nothing goes into a draft that your evidence bank can't support. The framework cannot — and must not — fabricate achievements, inflate scope, or claim outcomes that weren't yours. The watch-outs register exists specifically to catch overclaim before it reaches a page.
2. You review every word. There is a clear and intentional expectation that you read the final document end to end before it goes anywhere — checking that it is consistent, accurate, and truthful, and that it genuinely sounds like you. The framework drafts and verifies; you are the author and the accountable signatory. It is a co-pilot, never an autopilot.
3. The real value is the loop, not the document. When you review a draft and find something off — a weak framing, a stale metric, a claim that doesn't quite hold — you don't just fix the document. You fix the evidence bank. That correction makes the next application better, and the one after that. The compounding asset is the bank: a sharper, truer, better-organised account of your career that improves every time you use it. Document production is the by-product; continuous iteration on your evidence is the product.

If you want a button that fires off applications without your judgement in the loop, this is the wrong tool. If you want to get systematically better at representing your real, hard-won experience — application after application — this is built for exactly that.

What it does

Two workflows, one evidence base:

• Application drafting — drop a job URL or attachment into a Claude project and get a tailored cover letter, statement of claims, pitch, or selection-criteria response in the candidate's own voice, built from a structured evidence bank, gated by length and overclaim checks before delivery.
• Interview preparation — recruiter screens, behavioural panels, and executive panels: eight-section prep documents with STAR responses written for spoken delivery, honest timing audits at the candidate's real speaking pace, and capability mapping for framework-scored (e.g. APS SES) panels.

The machinery underneath:

• Evidence bank — a structured, tagged library of the candidate's proof points, positioning blocks, and ready-made templates, with a master index and per-role-family decision matrices that prescribe exactly which evidence stack fits which role archetype.
• Eleven verification skills (skills/) — mechanical pass/fail gates for overclaim (watch-outs sweep), length and platform character limits, matrix-row traceability, coverage analysis, STAR structure, speaking pace, capability mapping, pre-edit snapshots, file integrity, and personal-context privacy boundaries.
• Data-integrity disciplines — snapshot-before-edit, three-axis integrity checks, changelog splitting, and quarantine rules that keep a long-running, append-heavy workspace from silently corrupting.

Quick start

1. Clone or download this repository.
2. Copy the template/ folder into a fresh folder on your machine — this becomes your workspace.
3. Open a Cowork session (Claude desktop) pointed at that folder and paste the contents of template/01_SETUP_ORCHESTRATOR.md into the chat. Claude walks you through four setup phases: foundations, voice and style, evidence bank, and skill installation.
4. Paste the populated template/Project_Instructions.md into your Claude project's instructions field.
5. Drop a job ad URL into the project. The default action is a tailored cover letter, drawn from your evidence bank.
6. Read it end to end. Check it's accurate, consistent, and sounds like you. Where something is off, correct it — and push the correction back into the evidence bank so the next application starts stronger. This review step is not optional; it is the framework working as intended.

Repository layout

Path	What it is
template/	The de-personalised framework: project instructions, setup orchestrator, evidence-bank skeletons, voice-guide templates, changelog scaffolds
template/skills/	Eleven de-personalised skill skeletons (SKILL.md + references) — populate, package as .skill, and install per Setup Phase 4
template/Template_Changelog.md	Version history of the framework package (current: v0.5.6)
LICENSE	MIT
CONTRIBUTING.md	How to propose improvements

Requirements

• Claude desktop app with Cowork mode (for setup and file-producing sessions), or any Claude project for drafting-only use.
• For PDF interview-prep output: pandoc with xelatex (optional — Markdown output works without it).

Design principles

1. Evidence before prose. Nothing is claimed that the bank can't support; the watch-outs register records every framing that once went too far.
2. Ordering is sacred. Duplicate-check → gating → initiation → drafting → verification → maintenance. Skills implement content; the project instructions own the sequence.
3. Gates, not vibes. The checks most often skipped under pressure are the ones enforced mechanically.
4. Everything leaves a trail. Changelogs for the bank, the instructions, and the template itself; snapshots before structural edits; integrity checks before close.
5. Human in the loop, always. The framework drafts and verifies; you review, correct, and sign. The asset that compounds is your evidence bank, not any single document.

Provenance and licence

Extracted and de-personalised from a live framework maintained by Mikhail Dudarenok. MIT licence — use it, adapt it, ship your own version. Attribution appreciated, not required.