Roadmap beyond skills + MCP?

[not-stbenjam]

This is great to see. A vendor-neutral packaging format is badly needed, and skills + MCP was clearly the right minimal core to ship first. I maintain skillsaw, an open-source linter for agentic context files. Implementing several vendors' formats side by side left me with some questions about where this is heading.

What's on the roadmap past v1.0.0?

FUTURE_CONSIDERATIONS.md covers the cross-cutting concerns thoroughly but I didn't see anything about additional component types. Is that a deliberate boundary, in the sense that the spec models packaging rather than components? Or just not discussed yet?

On .agents/

.agents/ reads like the shared namespace. .agents/skills/ is honored across tools, and that's exactly the convergence everyone wanted. But Codex puts its marketplace catalog at .agents/plugins/marketplace.json, and that file carries a distinctly proprietary schema. So a vendor-specific format currently sits in what otherwise looks like commons.

Has there been any discussion about bringing a catalog format into the spec, or about what does and doesn't belong under .agents/? I ask partly as a tooling author, since anything reading that directory has to infer the ecosystem from the path, which quietly assumes no second vendor will put a differently-shaped file there.

Other primitives

Plugin packages in the wild already carry more than skills + MCP. I counted the official Codex catalog (180 plugins): 154 ship .app.json, 72 ship skills/, 14 ship plugin-level agents/, 6 ship commands/, and 2 ship hooks.json. Claude Code plugins similarly ship commands, agents, hooks, and path-scoped rules.

• Sub-agents, or packaged agent definitions. Multiple vendors ship these and there's no portable shape for them.
• Commands, same story - but favoring user-invocable skills is probably the right thing to do long term.
• Path-scoped rules and instructions, meaning conditional guidance like "when editing **/*.tf, do X". Probably the least standardized of the lot, with the highest value. Monolithic AGENTS.md are out of style IMHO.
• LSP integration. Agents burn tokens grepping for what a language server already knows precisely.
• Marketplaces and catalogs, per above.
• Hooks. I see the Hookbridge thread already exploring interop

Which of these are plausible candidates for a future version, which are deliberately out of scope, and would proposals be welcome?