Targets
A Gem is a neutral source. Targets turn it into something runnable — a generated project or an editor-ready config layout. Because every target reads the same Gem, you select where it goes without rebuilding from your raw config.
The targets
| Target | Kind | What it produces |
|---|---|---|
| Eve | Code-gen | A generated agent project from the Gem — the reference target pattern. |
| Flue | Code-gen / materialize | Materializes the Gem via the reusable compose hook into a Workers-style project. |
| OpenAI Sandbox | Code-gen | A SandboxAgent project with native stdio MCP, reusing the compose hook. |
| Bedrock AgentCore | Code-gen | An AgentCore harness project rendered from the Gem. |
| A2A | Code-gen / export | An A2A Agent Card, or a runnable agent-to-agent server (AI SDK v7) other agents can discover and call. |
Editor targets — Claude, Codex, Agents, Hermes, Cline/Roo, Gemini CLI, Continue, Cursor — render the Gem's artifacts into that ecosystem's config conventions.
All code-gen targets share a common compose step, so adding a new target mostly means
describing how it materializes a Gem — the introspection, selection, and redaction in
front of it are unchanged.
The A2A target is a bit different: rather than emitting a project to run, it exports your Gem as a discoverable Agent Card or a self-contained server for the agent-to-agent protocol.
From render to running
- Build a Gem (see Getting started).
- Pick a target. Materialize renders the project or config layout locally.
- Run it. A rendered Eve/Flue app can be started locally straight from the Materialize panel (see Testbed & run); other rendered projects are yours to run wherever you run Node projects.
Secrets are never written into a rendered project — they are referenced by name only (see Redaction).