proof-asset-creator

Overview

Help the user figure out what proof-of-value asset to build, tailored to their target roles, experience, and technical ability — then save a brief they can hand off. A resume says you can do the job; a proof asset shows it.

This skill is an ideation partner, not a producer. The hard part — and the only part this skill does — is matching what the user already has against what their target roles need against what they can actually ship. Once that match is decided and a brief is saved, the skill is done.

Drafting the actual case study, building the demo, or designing the portfolio piece is a job for a fresh agent conversation, where it has full context-window flexibility instead of being constrained by a skill workflow. Hand off the brief and continue there.

Workflow

State layer: reads source work documents (resume.md/cv.md), story-bank.md, existing proof-assets/, and applications.md. Writes one file to my-documents/proof-assets/{slug}.md. No tracker touch, no report. See state-layer contract.

1. Read the evidence layer

Before suggesting anything, read in priority order (see state-layer §8):

1. my-documents/resume.md or my-documents/cv.md — for experience and technical capability
2. my-documents/story-bank.md — STAR+R stories that could expand into case studies
3. my-documents/proof-assets/ — what the user already has (avoid duplicates, find gaps)
4. my-documents/applications.md — target roles the user is actively pursuing

Applications.md fallback: if the tracker is empty, thin, or the roles don't point in a clear direction, ask the user what kind of roles they're targeting. Don't make them opt in — just read it, and fall through to asking only when the data isn't there.

If source work documents are missing entirely, warn the user and offer resume-builder first. The skill still works on paste-only input, but evidence verification will be limited and every concrete claim will need [ASK: ...].