diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml
index e15ca79..7c1614b 100644
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@@ -4,7 +4,7 @@
 # file publish.yml. From then on, publishing a GitHub release ships to npm.
 #
 # Release channels map to npm dist-tags:
-#   latest  stable releases, what `npm install only-cli` gets
+#   latest  stable releases, what `npm install @only-cli/oc` gets
 #   beta    release candidates, version like 0.2.0-beta.1
 #   alpha   earlier previews, version like 0.2.0-alpha.1
 #   dev     throwaway builds from main, version stamped per run
diff --git a/README.md b/README.md
index 2f5bddf..aa70180 100644
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@ No per-site adapters required, no browser extension, no daemon. One generic dist
 ## Install
 
 ```
-npm install -g only-cli
+npm install -g @only-cli/oc
 ```
 
 Requires Node 20+. Requests go through [impers](https://github.com/lexiforest/impers) impersonating Chrome; if impers is unavailable the tool falls back to native fetch.
@@ -29,7 +29,7 @@ Requires Node 20+. Requests go through [impers](https://github.com/lexiforest/im
 
 The fastest setup is one line in your agent's instructions file (CLAUDE.md, AGENTS.md, or equivalent):
 
-> When you need content from a web page, run `npx only-cli open <url>` instead of fetching raw HTML. Run `npx only-cli --help` once to learn the commands.
+> When you need content from a web page, run `npx @only-cli/oc open <url>` instead of fetching raw HTML. Run `npx @only-cli/oc --help` once to learn the commands.
 
 Claude Code users can install the skill instead: copy `skills/only-cli/` into your project's `.claude/skills/` directory (or `~/.claude/skills/` to enable it everywhere). A skill costs almost no tokens until the agent actually invokes it, which fits how this whole project thinks. The same skill also installs through the [skills.sh](https://skills.sh) directory into Claude Code, Cursor, Codex, Copilot, and others:
 
@@ -37,7 +37,7 @@ Claude Code users can install the skill instead: copy `skills/only-cli/` into yo
 npx skills add only-cli/only-cli
 ```
 
-No setup at all also works: `npx only-cli` runs without a global install, and the tool teaches its own command surface through `--help`, the `actions:` line at the bottom of every render, and error messages that name the next command to run.
+No setup at all also works: `npx @only-cli/oc` runs without a global install, and the tool teaches its own command surface through `--help`, the `actions:` line at the bottom of every render, and error messages that name the next command to run.
 
 ## Commands
 
diff --git a/llms.txt b/llms.txt
index 924b755..d7c7c14 100644
--- a/llms.txt
+++ b/llms.txt
@@ -1,10 +1,10 @@
 # only-cli
 
-> only-cli (command: `oc`) turns websites into a command line interface for AI agents. Instead of fetching raw HTML (tens of thousands of tokens) or driving a headless browser with screenshots, an agent runs `npx only-cli open <url>` and gets a distilled text view of the page in a few hundred tokens, with numbered links and form fields it can act on. Works with Claude Code, Claude, ChatGPT, Codex, Gemini, Antigravity, and any agent that can run shell commands.
+> only-cli (command: `oc`) turns websites into a command line interface for AI agents. Instead of fetching raw HTML (tens of thousands of tokens) or driving a headless browser with screenshots, an agent runs `npx @only-cli/oc open <url>` and gets a distilled text view of the page in a few hundred tokens, with numbered links and form fields it can act on. Works with Claude Code, Claude, ChatGPT, Codex, Gemini, Antigravity, and any agent that can run shell commands.
 
 Key facts:
 
-- Install: `npm install -g only-cli`, or zero-install with `npx only-cli`
+- Install: `npm install -g @only-cli/oc`, or zero-install with `npx @only-cli/oc`
 - Commands: `oc open <url>` (compact view with numbered actions), `oc raw <url>` (whole page as markdown, `--html` for cleaned HTML), `oc --help` for the full surface
 - Default output budget is 500 tokens per page; `--budget <n>` adjusts it
 - Benchmarked at roughly 45x fewer tokens than reading raw HTML, with per-task numbers at https://github.com/only-cli/benchmarks
diff --git a/package.json b/package.json
index 11b53a3..483e70d 100644
--- a/package.json
+++ b/package.json
@@ -1,5 +1,5 @@
 {
-  "name": "only-cli",
+  "name": "@only-cli/oc",
   "version": "0.1.0",
   "description": "Turn websites into a compact CLI so AI agents can browse without burning tokens.",
   "type": "module",
diff --git a/skills/only-cli/SKILL.md b/skills/only-cli/SKILL.md
index 3a38aa6..c222fd9 100644
--- a/skills/only-cli/SKILL.md
+++ b/skills/only-cli/SKILL.md
@@ -10,8 +10,8 @@ Turns a web page into a compact terminal view instead of a raw HTML dump. A typi
 No install needed, run it with npx:
 
 ```
-npx only-cli open <url>     compact view with numbered elements
-npx only-cli raw <url>      whole page as markdown (add --html for cleaned HTML)
+npx @only-cli/oc open <url>     compact view with numbered elements
+npx @only-cli/oc raw <url>      whole page as markdown (add --html for cleaned HTML)
 ```
 
 ## Reading the output