Editorial pass on Skill Engineer docs

Tighten prose in the new Built-in agents and Skills picker
sections: replace internal jargon (allow-list, "binds the skills
toolset"), fix "greyed" -> "disabled", normalize "prefilled" to
the project's "pre-filled" convention, and break a dense
parenthetical action list in skills-build into a bulleted list.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: github-actions[bot]

docs/toolhive/guides-ui/playground.mdx

@@ -114,24 +114,23 @@ you want a different system prompt or model default.
 ### Enable installed skills
 
 When you select the **Skill Engineer** agent, the composer toolbar shows a
-**Skills** picker next to the Agent / MCP Servers / Model selectors. Use it to
-toggle which [installed skills](./skills-manage.mdx) are visible to the agent in
-the current chat:
+**Skills** picker next to the **Agent**, **MCP Servers**, and **Model**
+selectors. Use it to toggle which [installed skills](./skills-manage.mdx) are
+visible to the agent in the current chat:
 
 - The dropdown lists every skill on the **Installed** tab, deduplicated by name
-  and annotated with **User** or `<PROJECT_NAME>` badges that show each install
+  and tagged with a **User** badge or the project name to show each install
   location.
-- A live count badge on the trigger reflects how many skills are currently
-  enabled.
+- A count badge on the trigger shows how many skills are currently enabled.
 - Selecting a skill exposes its `SKILL.md` and bundled files to the agent so it
   can list, load, and read the contents on disk for an audit.
-- **Clear enabled skills** wipes the allow-list. Uninstalling a skill from the
-  **Skills** page also removes it automatically the next time the picker
-  refreshes.
+- **Clear enabled skills** disables every skill at once. Uninstalling a skill
+  from the **Skills** page also removes it automatically the next time the
+  picker refreshes.
 
-The picker is greyed out for agents that don't bind the skills toolset (such as
-the default ToolHive Assistant). Switch to **Skill Engineer** or any custom
-agent that binds the skills tool bundle to enable it.
+The picker is disabled for agents that don't have access to skill tools, such as
+the default ToolHive Assistant. Switch to **Skill Engineer** or a custom agent
+configured with the skills toolset to enable it.
 
 ## Using the playground
 

docs/toolhive/guides-ui/skills-browse-install.mdx

@@ -42,12 +42,12 @@ The **Install skill** dialog asks for:
     `ghcr.io/<OWNER>/skills/<SKILL_NAME>:<TAG>`).
   - A Git URL of the form `git://host/owner/repo[@ref][#path/to/skill]`.
 
-- **Version** (optional) - prefilled automatically when you open the dialog from
-  a registry skill or local build. If you paste a tagged or digest-pinned
+- **Version** (optional) - pre-filled automatically when you open the dialog
+  from a registry skill or local build. If you paste a tagged or digest-pinned
   reference into the **Reference** field (for example, `my-skill:v1.0.0` or
   `ghcr.io/<OWNER>/skills/<NAME>@sha256:<DIGEST>`), ToolHive splits the suffix
-  into the **Version** field when you tab out, so you don't have to break the
-  reference up by hand. Leave the field blank to install the latest version.
+  into the **Version** field when you tab out, so you don't have to split it
+  manually. Leave the field blank to install the latest version.
 
 - **Scope** (required) - where the skill files are written on disk. Hover the
   **(i)** icon next to the field to see the inline tooltip:

docs/toolhive/guides-ui/skills-build.mdx

@@ -68,14 +68,19 @@ that digest is removed.
 ## Build a skill from the Playground
 
 If you'd rather draft a skill conversationally, switch the
-[Playground](./playground.mdx) agent to **Skill Engineer**. The agent guides you
-through naming the skill, drafts the `SKILL.md` and any supporting files inside
-the chat, and then calls ToolHive's local build API to produce an OCI artifact.
+[Playground](./playground.mdx#built-in-agents) agent to **Skill Engineer**. The
+agent guides you through naming the skill, drafts the `SKILL.md` and any
+supporting files inside the chat, and then calls ToolHive's local build API to
+produce an OCI artifact.
 
 A successful build renders a "Skill built" card inline in the chat with three
-actions: **Install** (opens the **Install skill** dialog with the reference and
-version pre-filled), **View details** (deep-links to the build on the **Local
-Builds** tab), and **Copy name** (copies the bare skill name to the clipboard).
+actions:
+
+- **Install**: opens the **Install skill** dialog with the reference and version
+  pre-filled.
+- **View details**: opens the build's row on the **Local Builds** tab.
+- **Copy name**: copies the skill name to the clipboard.
+
 From there, the install and removal flows are identical to a build created from
 the **Build skill** dialog.
 

docs/toolhive/guides-ui/skills-manage.mdx

@@ -51,8 +51,8 @@ To inspect a skill's `SKILL.md` and bundled files without leaving the desktop
 app, switch the [Playground](./playground.mdx#built-in-agents) agent to **Skill
 Engineer**, enable the skill in the composer toolbar's **Skills** picker, and
 ask the agent to audit it. The agent reads the on-disk install, cross-checks
-every file the body references against what's actually packaged, and returns a
-PASS / PARTIAL / FAIL verdict.
+every file referenced in the `SKILL.md` body against the files packaged with the
+skill, and returns a PASS / PARTIAL / FAIL verdict.
 
 ## Next steps