fix(skill): set suggested model from original agent (#2371)

- Add Anti-Pattern bullet to llm-services.md: use the same model identifier
  as the original agent, not a generic one
- Add Step 5a checklist item to SKILL.md: set suggested model to match the
  model used in the original agent

Signed-off-by: penge <ppenge@gmail.com>

Author: penge

skills/agentstack-wrapper/SKILL.md

@@ -111,6 +111,7 @@ Task Progress:
 - [ ] Step 3: Add and Install Dependencies
 - [ ] Step 4: Create the Server Wrapper & Entrypoint
 - [ ] Step 5: Wire LLM / Services via Extensions (requires reading docs)
+- [ ] Step 5a: Set suggested model to match the model used in the original agent
 - [ ] Step 6: Implement Error Handling (requires reading docs)
 - [ ] Step 7: Map Forms (if applicable) (requires reading docs)
 - [ ] Step 7b: Adapt File Inputs (if applicable) (requires reading docs)
@@ -179,6 +180,8 @@ This classification determines:
 
 **Read [references/llm-services.md](reference/llm-services.md) and follow it completely for Step 5.**
 
+Set the `suggested` model in `LLMServiceExtensionSpec.single_demand()` to the same model used in the original agent.
+
 ---
 
 ## Step 6 – Error Handling

skills/agentstack-wrapper/references/llm-services.md

@@ -24,3 +24,4 @@ See the [chat agent](https://github.com/i-am-bee/agentstack/blob/main/agents/cha
 - Do not skip the docs/examples and improvise imports or extension usage.
 - Do not read model/API key/base URL from env vars when LLM extension is available.
 - Do not rewrite `api_base` heuristically unless the official docs explicitly require it.
+- Do not use a generic model as `suggested` — use the same model identifier as the original agent.