docs: fix broken ralph new check/context commands that no longer exist

The getting-started guide and README referenced `ralph new check <name>`,
`ralph new context <name>`, and `ralph new ralph <name>` — subcommands
that were removed in v0.1.8 when `ralph new` was redesigned as an
AI-guided session. New users following the getting-started tutorial
would hit errors at step 5.

Replaced with manual `mkdir` + file creation instructions, which are
predictable and copy-pasteable. Added a tip about `ralph new` as the
AI-guided alternative.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: 2 people

README.md

@@ -115,10 +115,10 @@ The simple loop works, but ralphify's real power comes from three primitives tha
 Checks validate the agent's work after each iteration. When one fails, its output automatically feeds into the next iteration so the agent can fix its own mistakes.
 
 ```bash
-ralph new check tests
+mkdir -p .ralphify/checks/tests
 ```
 
-Edit `.ralphify/checks/tests/CHECK.md`:
+Create `.ralphify/checks/tests/CHECK.md`:
 
 ```markdown
 ---
@@ -143,10 +143,10 @@ You define what "valid" means. Ralphify feeds failures back automatically.
 Contexts inject fresh data into the prompt each iteration — git history, test status, anything a shell command can produce.
 
 ```bash
-ralph new context git-log
+mkdir -p .ralphify/contexts/git-log
 ```
 
-Edit `.ralphify/contexts/git-log/CONTEXT.md`:
+Create `.ralphify/contexts/git-log/CONTEXT.md`:
 
 ```markdown
 ---
@@ -162,8 +162,8 @@ The command runs before each iteration. Use `{{ contexts.git-log }}` in your `RA
 Keep multiple ralphs for different jobs and switch between them at run time:
 
 ```bash
-ralph new ralph docs
-ralph new ralph refactor
+ralph new docs
+ralph new refactor
 ```
 
 Edit `.ralphify/ralphs/docs/RALPH.md` with your documentation-focused prompt, then:

docs/getting-started.md

@@ -103,10 +103,10 @@ Checks run **after** each iteration to validate the agent's work. If a check fai
 Create a check that runs your test suite:
 
 ```bash
-ralph new check tests
+mkdir -p .ralphify/checks/tests
 ```
 
-This creates `.ralphify/checks/tests/CHECK.md`. Edit it:
+Create `.ralphify/checks/tests/CHECK.md`:
 
 ```markdown
 ---
@@ -125,10 +125,10 @@ The text below the frontmatter is the **failure instruction** — it gets includ
 Add a second check for linting:
 
 ```bash
-ralph new check lint
+mkdir -p .ralphify/checks/lint
 ```
 
-Edit `.ralphify/checks/lint/CHECK.md`:
+Create `.ralphify/checks/lint/CHECK.md`:
 
 ```markdown
 ---
@@ -144,10 +144,10 @@ Fix all lint errors. Do not suppress warnings with noqa comments.
 Contexts inject dynamic data into the prompt before each iteration. A useful default is recent git history — it helps the agent understand what's already been done.
 
 ```bash
-ralph new context git-log
+mkdir -p .ralphify/contexts/git-log
 ```
 
-Edit `.ralphify/contexts/git-log/CONTEXT.md`:
+Create `.ralphify/contexts/git-log/CONTEXT.md`:
 
 ```markdown
 ---
@@ -160,6 +160,9 @@ enabled: true
 
 The command runs each iteration and its output is appended to the prompt. The body text ("## Recent commits") appears above the command output as a label.
 
+!!! tip "Use `ralph new` for AI-guided setup"
+    Instead of creating primitive files manually, you can run `ralph new` to launch an AI-guided session that creates a complete ralph — prompt, checks, and contexts — via conversation with your agent.
+
 ### Declare your primitives and place the context
 
 Checks and contexts must be **declared** in your ralph file's frontmatter before they'll run. Now that you've created all three primitives, update `RALPH.md` to declare them and place the context: