Drop dashboard refresh from the Drafty plugin

README.md

@@ -18,10 +18,9 @@ Or hand the whole thing to your agent — paste this into Claude Code: *"Set me
 - **"drafty it."** Tell Claude to drafty the thing it just wrote → you get a `drafty.im/canvas/<slug>` link. Anyone you share it with hovers an element, clicks, and leaves a threaded comment — live cursors, no sign-up needed to comment.
 - **Claude closes the loop.** "Address the canvas" → Claude reads each thread, edits the source file, pushes a new version on the same link, replies on the canvas, and marks threads done. Set a canvas `live` and Claude works comments as they arrive.
 - **Agent eyes.** `drafty shot` renders a canvas, a local HTML file, or any URL to an image so Claude can *see* what it built — including a commenter's exact view (their viewport width, their revision, the anchored element highlighted) instead of guessing at "looks squished on my phone" from text alone.
-- **Site boards.** `drafty present <url>` maps a site (robots → sitemap → homepage links), curates the main screens, captures each at desktop + phone width with local Chrome, and publishes an annotatable board. `--refresh` re-shoots the same screens as a tick — competitor tracking, staging watch.
+- **Site boards.** `drafty present <url>` maps a site (robots → sitemap → homepage links), curates the main screens, captures each at desktop + phone width with local Chrome, and publishes an annotatable board for review.
 - **Versioned, with a real undo.** Every push snapshots a revision. `revert` rolls the canvas back AND resyncs the local file atomically; `status` reports in-sync / local-ahead / canvas-ahead / diverged; a push that would clobber an edit made elsewhere (browser, another agent) is refused with instructions, never silently applied.
 - **Organized, and self-tidying.** Projects, tags, pin, archive — and `drafty tidy --sweep` cross-references your canvases against the repo's git log to flag the ones whose work already shipped, so finished specs get receipted and archived instead of rotting on the list.
-- **Self-refreshing dashboards.** The bundled `drafty-cron` skill wires a plain OS cron — no model, no credits at runtime — that re-renders and pushes a data-backed canvas on a timer.
 
 ## How it works
 
@@ -37,7 +36,6 @@ You talk; Claude runs the commands. You never touch the CLI yourself.
 
 - **The `drafty` skill** — teaches Claude the whole loop: publish, read comments, reply, mark threads done, push revisions, render-and-look before claiming a visual fix, sweep shipped canvases, roll back. Claude loads it on its own when you say "drafty it" / "share this for feedback" / "what did they comment".
 - **The `drafty` CLI** — a single-file, thin HTTP client. Every command is a call to drafty.im's public API; ownership and visibility are enforced server-side. No keys ship in the plugin — auth is a browser sign-in (`drafty login`), one sign-in covering web + CLI.
-- **The `drafty-cron` skill** — the control plane for self-refreshing canvases (author a deterministic query → render → push script once; launchd runs it forever).
 
 ## Modes & visibility
 
@@ -88,7 +86,7 @@ Claude drives these; the reference is here so you can audit what it's doing.
 | Command | What it does |
 |---|---|
 | `drafty shot <slug\|file.html\|url> [--width N] [--annotation A] [--full]` | Render to an image, print the path. `--annotation` reproduces a commenter's exact view. Local files/URLs and private canvases render with your own headless Chrome; public canvases use the server's cached render. |
-| `drafty present <url> [--screens N] [--urls …] [--slug S --refresh] [--dry-run]` | Site board: map → curate (≤20 screens) → shoot at 1280+390px → publish annotatable board. |
+| `drafty present <url> [--screens N] [--urls …] [--slug S] [--dry-run]` | Site board: map → curate (≤20 screens) → shoot at 1280+390px → publish an annotatable snapshot. Reuse `--slug` to publish a new version on the same canvas. |
 
 **Session**