blog: Code Share — Deploy kagent with Agent Substrate

[sebbycorp]

Add a hands-on guide by Sebastian Maniak for running kagent OSS wired to Agent Substrate on a kind cluster: the concepts (actors, workers, snapshots, suspend/resume), use cases, a one-shot setup path, and a manual Helm walkthrough. Includes kagent UI screenshots of the Agents view and the View → Substrate page, and points to the runnable code at sebbycorp/kagent-demos.

[Copilot]

Pull request overview

Adds a new “Code Share” blog post describing how to deploy kagent OSS with Agent Substrate on a kind cluster, including concepts, use cases, and step-by-step setup (scripted and manual), with supporting screenshots.

Changes:

• Adds a new MDX blog post with metadata, walkthrough, and troubleshooting section.
• Updates the blog index page to include the new post in the internal posts list.

Reviewed changes

Copilot reviewed 2 out of 4 changed files in this pull request and generated 1 comment.

File	Description
src/blogContent/deploy-kagent-with-agent-substrate.mdx	New long-form guide post (concepts + setup steps + screenshots/links).
src/app/blog/page.tsx	Adds the new post entry to the blog listing.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[artberger]

Left some formatting-related comments, but approved.

[artberger]

Suggested change

**[Agent Substrate](https://github.com/agent-substrate/substrate)** flips the model. It decouples the *agent session* from the *pod*: idle sessions are checkpointed — full RAM and filesystem, via [gVisor](https://gvisor.dev/) — to object storage, the pod returns to a warm pool, and the session resumes **sub-second** on the next request, exactly where it left off. Think serverless scale-to-zero, but for *stateful* agents.

**[Agent Substrate](https://github.com/agent-substrate/substrate)** flips the model. It decouples the *agent session* from the *pod*: idle sessions are saved to object storage (full RAM and filesystem, via [gVisor](https://gvisor.dev/)), the pod returns to a warm pool, and the session resumes **sub-second** on the next request, exactly where it left off. Think serverless scale-to-zero, but for *stateful* agents.

[artberger]

For some reason, the code ticks throughout still are rendered, probably some bug in the styling.

[artberger]

Adding this to the tailwind.config.ts file starting at L59 resolved the issue for me locally:

  		typography: {
  			DEFAULT: {
  				css: {
  					// The typography plugin wraps inline <code> in literal backtick
  					// characters via ::before/::after. We render our own styled
  					// <code>, so strip them.
  					'code::before': { content: '""' },
  					'code::after': { content: '""' }
  				}
  			}
  		}

[artberger]

The ASCII diagram doesn't render well:

[artberger]

In the src/components/mdx/code-block.tsx file starting at L25, this fixed the rendering:

    const content = typeof children === 'string' ? children : String(children);
    // A fenced code block without a language (e.g. ASCII diagrams) has no
    // className but spans multiple lines. Render it in a real <pre> so
    // whitespace and newlines are preserved instead of collapsing to inline.
    if (content.includes('\n')) {
      return (
        <pre className="overflow-x-auto p-4 text-sm rounded-lg bg-muted font-mono text-foreground whitespace-pre">
          <code className="font-mono">{content.replace(/\n$/, '')}</code>
        </pre>
      );
    }

[artberger]

Similar comment on this diagram not rendering, works with the fix above.

Before

After