Add a concise Agent Backend example

[lpc0387]

2. Add a concise Agent Backend example

Labels: good first issue, documentation, examples

Body:

The README explains that Claude Code, Hermes Agent, Codex, OpenCode, and custom command-backed agents can share the same Agent Backend abstraction. Add a small example document that shows the shape of a custom command backend configuration and the questions a contributor should answer before implementing one.

Suggested scope:

• Create a concise example under examples/ or a short section linked from README.en.md.
• Include a minimal command-backed backend sketch using placeholder values only.
• Explain what should be verified locally before proposing a backend.
• Avoid real credentials, local machine paths, or private model configuration.

Acceptance criteria:

• A newcomer can read the example and understand what a custom backend is responsible for: command invocation, working directory, timeout, and result capture.
• The example uses placeholders and does not imply that secrets belong in the repository.
• The README or contributor docs link to the example if a new file is added.
• Verification notes include a Markdown review and any narrow syntax check relevant to files changed.

Why it is newcomer-friendly:

The task is mostly explanatory writing. It introduces an important project concept while avoiding changes to backend execution code.

---

This issue was created from the v5.0 launch backlog in docs/launch/good-first-issues.md.