lavish-axi

Original：🇺🇸 English

Translated

Turn complex or visual agent responses into rich, reviewable HTML artifacts the user can annotate and send feedback on, using the lavish-axi CLI. Use when about to give a plan, comparison, diagram, table, code diff, report, or anything easier to grasp visually than as prose.

9installs

Sourceyjuyjuy/lavish-axi

Added on2026-09-03

NPX Install

npx skill4agent add yjuyjuy/lavish-axi lavish-axi

SKILL.md Content

View Translation Comparison →

lavish-axi

Open an agent-generated HTML artifact in the browser as a review surface a human can annotate, then read that feedback back over a long poll.

When to reach for it

A plan, comparison, diagram, table, code view, report, prototype, or decision surface will be clearer as a page than as prose: build the HTML, then open it here.
Structured input from the user (choices, triage, scope) belongs in an artifact with the
```
input
```
playbook, not in a chat question.
Do NOT reach for it for a short textual answer, or when nobody is at a browser; answer in the conversation instead.
Sharing outside this machine is
```
export
```
(a portable file) or
```
share
```
(a hosted page), not the review server.

Workflows

Build the artifact only after reading the current guidance; it is CLI-owned and installed skills go stale.

bash

lavish-axi design                  # design-direction priority, CDN snippet, playbook router
lavish-axi playbook                # list playbook ids
lavish-axi playbook plan           # focused guidance for the artifact you are about to write

Review loop - open, then wait for the human. The poll is silent by design; never kill it.

bash

lavish-axi .lavish/report.html                       # prints the session url
lavish-axi poll .lavish/report.html                  # blocks until feedback or session end
lavish-axi poll .lavish/report.html --agent-reply "fixed the pricing table"

Hand the artifact to someone who is not on this box.

bash

lavish-axi export .lavish/report.html --out /tmp/report-standalone.html
lavish-axi share .lavish/report.html --password "<pw>"   # third-party host, public without --password

Close out or recover.

bash

lavish-axi                          # live sessions, or an empty list
lavish-axi end .lavish/report.html  # agent-side end; a later plain open still works
lavish-axi stop                     # shut the background server down

Conventions

Write artifacts under

.lavish/

in the working directory unless the user names a place. Keep sibling assets in that same directory and reference them with relative paths; a leading

does not resolve.

Every response ends with

next_step

or

help:

hints; follow them rather than guessing the next command. Every command takes

--help

.

Detected layout issues never return the poll. They wait in the user's Layout issues inbox and arrive as an ordinary

layout-warnings

prompt only once the user queues them, so never edit the artifact to chase a layout issue the user did not send.

Run the poll in the foreground unless the harness has a tracked background job that provably wakes the same agent. Never

nohup

,

, or

disown

it. A killed poll loses nothing: re-run it.

A browser-side end refuses a later plain reopen; pass

--reopen

only when the user asks or something genuinely needs their eyes.

No global install required:

npx -y lavish-axi ...

runs every command above, and follow-up commands the CLI prints should be run that way too.

Non-goals

Not a design system reference:
```
lavish-axi design
```
and
```
lavish-axi playbook <id>
```
own that, and this file must never restate it.
Not a flag reference.
```
--help
```
is current; this skill is not.
```
share --unpublish
```
does not delete a hosted page, it overwrites it. The URL still resolves.
No reverse conversion: whiteboard edits are applied by updating the Mermaid source, never by writing a scene file back.

Request

$ARGUMENTS

If the request above is non-empty, the user invoked

/lavish-axi

explicitly - fetch the current CLI guidance, then build that artifact. If it is empty, infer what to visualize from the conversation.

lavish-axi

NPX Install

Tags

SKILL.md Content

lavish-axi

When to reach for it

Workflows

Conventions

Non-goals

Request