2026-11-02 baseline: 0 completed external pilots recorded. This procedure is not evidence of adoption or time saved. Stop after ten minutes, even if blocked.
With Node.js 22 or later:
npx --yes @hyscodebase/skillwhy@0.1.0-alpha.4 demoThe initial compiled-package download needs internet access. The synthetic demo needs no Codex, API key, or model call. Demo-only feedback is welcome and is kept separate from live native observations. Note your approximate time to first success.
Live capture supports Codex CLI 0.166.0 on macOS arm64 and Linux x64
(Linux native checks use Ubuntu 24.04). Check your existing
environment with codex --version; use node --version and
node -p "process.platform + '/' + process.arch" for the other runtime details.
If unsupported or absent, finish with demo feedback. Do not install, upgrade, or
downgrade Codex for this pilot.
If you do not already have SkillWhy installed, use the compiled release:
npm install --global @hyscodebase/skillwhy@0.1.0-alpha.4
skillwhy --versionUse an existing skill project. Replace ./my-project, then choose one returned ID:
skillwhy capture --cwd ./my-project --out pilot.snapshot.json
skillwhy inspect pilot.snapshot.json
# Replace ENTRY_ID with the exact ID of the one entry you chose:
skillwhy share --snapshot pilot.snapshot.json --entry ENTRY_ID --alias pilot-review --previewTo filter duplicate names, add --skill YOUR_SKILL_NAME to inspect; choose one
ID. An empty result is also feedback. Keep the snapshot and inspection local:
they contain private metadata. Use a new filename if an output already exists.
Capture requests fresh CLI metadata, not model turns or skill execution. Codex can still perform its own startup/cache/network activity. Preview uploads nothing.
If a second environment is available, each person selects their entry and
separately confirms that the entries correspond before choosing the same
neutral --alias. SkillWhy does not establish this association for you.
After reviewing both previews, replace --preview with --out local.share.json
and --out other.share.json respectively. Exchange only reviewed bundles, then:
skillwhy diff local.share.json other.share.json --detailThis compares selected observations, not model behavior. This step is optional.
Optionally publish a sanitized summary using the pilot report form: runtime versions, demo versus live attempt, time to first success, blocked step, whether it helped your problem, and whether you would use it again. Estimates and unsuccessful attempts are useful; no successful outcome is assumed.
Do not attach raw snapshots, configuration, skill bodies, private paths, API keys, or secrets. No attachment is needed. Nothing is sent automatically; filing an issue is your choice.
Maintainers can use the interview guide to record the problem, whether assistance was needed, and the participant's interpretation of the result. See troubleshooting for incomplete captures or comparisons.