docs: normalize preset docs and humanize prose (#690)

Apply consistent section structure across preset docs: add Skill Reference
to opencode-go, 3-column tables with Source everywhere, and a Configuration
link in each. Humanize the non-config prose.

docs/authors-preset.md

@@ -149,3 +149,5 @@ Each skill is listed with a short description and its source. The config block a
 | `websearch` | (MCP) web search | `public` |
 | `workers-best-practices` | Worker best practices | `author` |
 
+For the complete configuration reference, see [Configuration](configuration.md).
+

docs/openai-preset.md

@@ -1,7 +1,7 @@
 # OpenAI Preset
 
-`openai` is the bundled default generated preset. The installer generates both
-`openai` and `opencode-go` presets; OpenAI is active by default.
+`openai` is the default generated preset. The installer builds both `openai`
+and `opencode-go`; OpenAI is the one that runs unless you pick the other.
 
 ## Install with OpenAI Active
 
@@ -29,8 +29,7 @@ workflow.
 
 ## Bundled Model Mapping
 
-The generated `openai` preset maps each specialist to an OpenAI model tuned for
-its role:
+The generated `openai` preset assigns each specialist an OpenAI model:
 
 | Agent | Model |
 |-------|-------|

docs/opencode-go-preset.md

@@ -1,11 +1,10 @@
 # OpenCode Go Preset
 
-`opencode-go` is a bundled generated preset for users who want to run the
-Pantheon agents through OpenCode Go models instead of the default OpenAI setup.
+`opencode-go` runs the Pantheon agents on OpenCode Go models instead of the
+default OpenAI setup.
 
-The installer generates both `openai` and `opencode-go` presets. OpenAI stays
-active by default unless you select OpenCode Go during install or switch to it
-later.
+The installer builds both `openai` and `opencode-go`. OpenAI stays active unless
+you choose OpenCode Go at install time or switch to it later.
 
 Because the `opencode-go` preset uses GLM-5.1 for Orchestrator and GLM is not
 multimodal, installing with `--preset=opencode-go` also enables the Observer
@@ -67,13 +66,19 @@ setting the top-level `preset` field:
   "disabled_agents": [],
   "presets": {
     "opencode-go": {
-      "orchestrator": { "model": "opencode-go/glm-5.2" },
+      "orchestrator": {
+        "model": "opencode-go/glm-5.2"
+      },
       "oracle": {
         "model": "opencode-go/qwen3.7-max",
         "variant": "max"
       },
-      "librarian": { "model": "opencode-go/deepseek-v4-flash" },
-      "explorer": { "model": "opencode-go/deepseek-v4-flash" },
+      "librarian": {
+        "model": "opencode-go/deepseek-v4-flash"
+      },
+      "explorer": {
+        "model": "opencode-go/deepseek-v4-flash"
+      },
       "designer": {
         "model": "opencode-go/kimi-k2.7-code",
         "variant": "medium"
@@ -82,13 +87,21 @@ setting the top-level `preset` field:
         "model": "opencode-go/deepseek-v4-flash",
         "variant": "high"
       },
-      "observer": { "model": "opencode-go/kimi-k2.6" }
+      "observer": {
+        "model": "opencode-go/kimi-k2.6"
+      }
     }
   }
 }
 ```
 
-This preset defines no `skills` or `mcps` per agent. It relies on model selection only (see [Bundled Model Mapping](#bundled-model-mapping)); agents use whatever skills are globally installed.
+## Skill Reference
+
+This preset defines no per-agent `skills` or `mcps`. All agents use whatever skills are globally installed (the `*` wildcard).
+
+| Skill | Description | Source |
+| --- | --- | --- |
+| `*` | All installed skills (wildcard) | public |
 
 For the complete configuration reference, see
 [Configuration](configuration.md).

docs/thirty-dollars-preset.md

@@ -1,8 +1,8 @@
 # $30 Preset
 
-This preset is for people who want a strong setup built around **Codex Plus ($20/month)** and **GitHub Copilot Pro ($10/month)**.
+This preset pairs **Codex Plus ($20/month)** with **GitHub Copilot Pro ($10/month)**.
 
-It uses Codex Plus for the OpenAI models and GitHub Copilot for the premium design models, giving you a mixed-provider setup for about **$30/month total**.
+Codex Plus covers the OpenAI models and Copilot covers the design models, so you get a mixed-provider setup for about **$30/month total**.
 
 ---
 
@@ -25,6 +25,8 @@ It uses Codex Plus for the OpenAI models and GitHub Copilot for the premium desi
 
 ## Skill Reference
 
-| Skill | Description |
-| --- | --- |
-| `*` | All installed skills (wildcard) |
+| Skill | Description | Source |
+| --- | --- | --- |
+| `*` | All installed skills (wildcard) | public |
+
+For the complete configuration reference, see [Configuration](configuration.md).