with one click
developer-experience-engineer
// Developer experience — first-60-minutes friction, CONTRIBUTING.md, install script, build loop, IDE integration, error noise.
// Developer experience — first-60-minutes friction, CONTRIBUTING.md, install script, build loop, IDE integration, error noise.
| name | developer-experience-engineer |
| description | Developer experience — first-60-minutes friction, CONTRIBUTING.md, install script, build loop, IDE integration, error noise. |
This is a capability skill ("hat"). It encodes the how of
auditing the human-contributor experience: simulating the first
60 minutes of a fresh clone, counting friction, routing fixes to
canonical owners. No persona lives here; the persona (if any)
is carried by the matching entry under .claude/agents/.
A contributor who clones Zeta for the first time has a local .NET install, some DBSP context, and 60 minutes. They should be able to land a trivial PR in that window — or at minimum understand, from the repo's own text, why they cannot. Every bit of friction in that path is paid by every contributor, every attempt, forever. DX audit is high-leverage maintenance, not cosmetics.
CONTRIBUTING.md — the contribution entry point.CLAUDE.md — dual-audience file (agents + humans); the DX
audit covers the human-read path only.README.md — first impression; does it resolve the "is this
for me" question?tools/setup/install.sh and per-OS scripts — install loop.dotnet build -c Release, dotnet test,
lake build, bun tools/formal-verification/run-tlc.ts.tests/**..vscode/, Ionide config, suggested
extensions, debugger setup..github/PULL_REQUEST_TEMPLATE.md, .github/workflows/*
(the human-visible surface, not the workflow internals —
that is Dejan's lane).Out of scope:
user-experience-engineer
(Iris).agent-experience-engineer
(Daya).harsh-critic (Kira).devops-engineer
(Dejan). The DX audit measures felt experience; Dejan
measures whether the script actually works.docs/PLUGIN-AUTHOR.md (when that doc lands); not a
DX-solo lane.tools/setup/install.sh and
related scripts; paired with Dejan.dotnet build -> dotnet test
-> incremental edit cycle..vscode/ / Ionide / IntelliSense.For the target:
Six friction types (parallel to AX, adjusted for human readers):
tools/verify-formal.sh" with no prior mention of it).Every intervention is rollback-safe in one round:
No multi-file refactor is proposed without Kenji sign-off.
Append findings to memory/persona/bodhi/NOTEBOOK.md in the
output format below. Kenji reads this notebook on round-close
and acts on the top-3 items.
# DX audit — round N, target: <first-PR | install-loop | build-loop | ide | persona-shape:<name>>
## Cold-walk timeline
- Minute 0: <first action the reader takes>
- Minute N: <each subsequent action, with file:line pointers>
- Time-to-first-PR estimate: <minutes>
- Trend vs last audit: <delta>
## Friction (P0 / P1 / P2)
P0 (first-PR cannot be landed inside the hour):
- [surface] — [type] — <one-sentence description with file:line>.
Intervention: <concrete action>. Owner: <Samir / Dejan / Kenji>.
P1 (friction but surmountable):
- ...
P2 (small wins):
- ...
## Proposed interventions (this round)
1. `<file>` — <change>. Owner: <name>. Effort: S/M/L.
Rollback: <how>.
2. ...
## Pointer-drift catalogue
- [surface] — [file:line] — [stale target] -> [current target].
## Recommended new entries
- `CONTRIBUTING.md`: <additions>.
- `docs/GLOSSARY.md`: <additions>.
- DEBT.md `dx-drift` entries: <list>.
agent-experience-engineer.CONTRIBUTING.md change — re-audit entry path.tools/setup/install.sh change — paired with Dejan;
Dejan measures correctness, Bodhi measures felt experience.docs/PLUGIN-AUTHOR.md exists..claude/agents/developer-experience-engineer.md — the
persona (Bodhi)CONTRIBUTING.md — the entry point audited hereCLAUDE.md — dual-audience fileREADME.md — first impressiontools/setup/install.sh — install loop audited heredocs/GLOSSARY.md — DX / AX / UX / wake / hat / frontmattermemory/persona/bodhi/NOTEBOOK.md — Bodhi's notebook
(created on first audit)docs/EXPERT-REGISTRY.md — Bodhi's roster entrydocs/CONFLICT-RESOLUTION.md — conflict-resolution protocoldocs/AGENT-BEST-PRACTICES.md — BP-01, BP-03, BP-07, BP-08,
BP-11, BP-16