docs: lead with the skill install, drop the npm framing

CHANGELOG.md

@@ -8,7 +8,7 @@ Notable changes to jev-writer. Format loosely follows [Keep a Changelog](https:/
 - `jev-writer dashboard` renders `report.json` as one self-contained HTML page: inline CSS and JS, data embedded, no CDN and no build step, so it opens over `file://`. Exposed as `buildDashboard()` for programmatic use.
 - `corpus` facts on the report: posting volume by year, follower trajectory, and headline medians.
 - A derived posting-cadence classification, so the dashboard writes a true sentence about your volume instead of a templated one.
-- An `exports` map, making `jev-writer/rubrics/pack.mjs`, `jev-writer/stats.mjs`, `jev-writer/report.mjs`, `jev-writer/dashboard.mjs` and `jev-writer/adapters/linkedin.mjs` the stable public surface.
+- An `exports` map for library consumers, making `jev-writer/rubrics/pack.mjs`, `jev-writer/stats.mjs`, `jev-writer/report.mjs`, `jev-writer/dashboard.mjs` and `jev-writer/adapters/linkedin.mjs` the stable public surface.
 - Question groups on the LinkedIn pack, covering all 30 questions across 5 families. Presentation only.
 - CI on Node 20 and 22, plus a smoke job validating every rubric pack without a key.
 - Tests for `report.mjs` and `dashboard.mjs`, which had none. Suite grew from 27 to 67.

README.md

@@ -17,6 +17,14 @@ jev-writer rates every post you have published against a pre-registered rubric,
 
 It analyses posts. It does not write them.
 
+Install it as an agent skill and ask your agent to analyse your writing:
+
+```bash
+npx skills add Kaos599/jev-writer --skill jev-writer
+```
+
+The skill walks you through exporting your data, getting a key, running the pipeline, and reading the result without overstating it. Or drive the CLI yourself:
+
 ```bash
 npx github:Kaos599/jev-writer doctor          # check your key, make one live call
 npx github:Kaos599/jev-writer run ./exports   # build, rate, analyse, render
@@ -331,8 +339,8 @@ Because you have fewer than 25 posts with outcome data. At that size, correlatio
 **Can I use a normal LLM instead of Jev?**
 Partly. The AI SDK's `evaluate` works with OpenAI, Anthropic, and Google models through an adapter, but those do not return calibrated probability distributions, so the rubric-health check and the confidence gating stop working.
 
-**Is it on npm?**
-Not yet. `npx github:Kaos599/jev-writer` installs straight from this repo and works today.
+**How do I install it?**
+Either way works and neither needs npm. `npx skills add Kaos599/jev-writer --skill jev-writer` installs the agent skill into Claude Code, Codex, and anything else the skills CLI supports. `npx github:Kaos599/jev-writer` runs the CLI straight from this repo.
 
 ## Contributing
 

package.json

@@ -69,8 +69,5 @@
     "./adapters/linkedin.mjs": "./src/adapters/linkedin.mjs",
     "./package.json": "./package.json"
   },
-  "author": "Kaos599",
-  "publishConfig": {
-    "access": "public"
-  }
+  "author": "Kaos599"
 }