bug: Amazon Q CLI lacks persistent custom instruction functionality

### Checks

- [x] I have searched [github.com/aws/amazon-q-developer-cli/issues](https://github.com/aws/amazon-q-developer-cli/issues?q=) and there are no duplicates of my issue
- [x] I have run `q doctor` in the affected terminal session
- [x] I have run `q restart` and replicated the issue again

### Operating system

MacOS Sequoia

### Expected behaviour

**Expected Behavior:**
Users should be able to set persistent custom instructions (like system prompts) that automatically apply to every Q CLI session, similar to ChatGPT's custom instructions feature. This should work reliably without complex workarounds.

**Impact:**
This is a fundamental UX limitation that prevents users from customizing Q CLI behavior for their workflows. Other AI tools provide this basic functionality. Without this feature, Q CLI is significantly less useful for professional development work.

**Suggested Solution:**
Add a command like `q settings instructions set "custom instructions here"` or similar functionality to persist user instructions across sessions.

**Developers Notes**
So much potential for q but simple functionality like this shows there is no care for a good user experience and was very disapointing to see I had to create an issue for something so basic and simple. If this cant be added, I cant see ever paying for monthly subscription. Currently im just having to do `/load` and direct q to a file that contains how i want it to work, that is an **INSANE** way to do things *every single time* I start up a new instance of q chat. PLEASE improve this. **User experience is very very important**

### Actual behaviour

1. Custom agents fail with tool validation errors even with minimal configurations
2. No built-in way to set persistent custom instructions
3. /load command expects conversation format, not instruction format
4. /context command only handles file context rules, not system instructions
5. Piping instructions as first message treats each instruction as a separate question to answer
6. Users must manually copy/paste instructions every session or use unreliable workarounds

### Steps to reproduce

1. Try to create a custom agent with basic instructions - fails with tool validation errors
2. Try to use /load with instruction file - fails with format errors
3. Try to use /context for system instructions - only works for file context
4. Try piping instructions as first message - Q responds to each instruction individually instead of treating them as system context
5. Result: No reliable way to set persistent custom instructions

### Environment

```yaml
q cli needs **heavy work** done to improve… even running q diagnostic doesnt give you actual diagnostic data, only a menu, here were the results after running q diagnostic:
> q diagnostic

🛠️ Using tool: introspect (trusted)
⋮
● Introspecting to get you the right information ⋮
● Completed in 0.0s

⠋ Thinking...
> I don't see a /diagnostic command in the Q CLI documentation. The available commands are:

• /quit - Quit the application
• /clear - Clear conversation history
• /agent - Manage agents
• /context - Manage context files
• /editor - Open editor to compose prompts
• /reply - Open editor with quoted assistant message
• /compact - Summarize conversation
• /tools - View tools and permissions
• /issue - Create GitHub issue
• /logdump - Create zip file with logs for support
• /changelog - View Q CLI changelog
• /prompts - View and retrieve prompts
• /hooks - View context hooks
• /usage - Show context window usage
• /mcp - See loaded MCP servers
• /model - Select conversation model
• /experiment - Toggle experimental features
• /subscribe - Upgrade to Q Developer Pro
• /save - Save current conversation
• /load - Load previous conversation
• /todos - View and manage TODO lists
• /paste - Paste image from clipboard
• /help - Print help message

If you're looking for diagnostic information, you might want:
• /logdump - Creates a zip file with logs for support investigation
• /tools - Shows available tools and their permissions
• /usage - Shows current context window usage

Could you clarify what diagnostic information you're looking for?

Reference: https://github.com/aws/amazon-q-developer-cli/blob/main/docs/built-in-tools.md
```

Commencez par les points d’entrée de la CLI pour /load, /context et /agent, puis lisez docs/built-in-tools.md afin de comprendre le comportement existant des sessions et du contexte. Définissez comment les instructions persistent entre les sessions et comment l’achèvement peut être vérifié, tout en séparant la demande de custom-instructions des problèmes de diagnostic et de custom-agent signalés.

Rédigé par le modèle d'indexation à partir du texte de l'issue.