Skip to content

Community Roadmap

The book already has a strong core: architecture, safety, memory, execution, observability, organizational design, and a runnable reference package. So what follows is not a general wishlist, but the backlog for the next layer of value.

What Is Already Done

How I Select the Next Steps

Each item below passes three tests:

  • it can be applied quickly in real work;
  • it helps the wider community, not only the author of the book;
  • it improves international readability, not only technical depth.

Next 10 Improvements

1. Extend the case studies set

The current cases are already useful, but the book should add 2-3 more:

  • enterprise workflow agent;
  • CRM/task agent;
  • security-sensitive assistant.

Why it matters: the more recognizable scenarios readers see, the easier it is to map the architecture to their own systems.

2. Extend the policy template set

The book already contains first examples, but the community needs more explicit templates:

  • tool approval policy;
  • memory write policy;
  • egress policy;
  • rollout gate policy;
  • retrieval policy.

Why it matters: a template people can copy and adapt is more valuable than a long explanation of why the template matters.

3. Richer trace examples and visual QA

The trace schema and event catalog exist already. The next step is to make them easier to validate and read:

  • add more realistic JSONL trace examples;
  • show one successful run, one approval wait, one denied run, and one failed run;
  • keep the rendered schema pages free of broken tables, list joins, and raw diagram blocks;
  • add a lightweight rendered-site QA checklist for the most visible pages.

Why it matters: observability becomes much stronger when the community has both a shared event model and concrete examples of how it appears in the published site.

4. More realistic scenarios in agent_runtime_ref

The package is already useful, but the next step should add:

  • one knowledge scenario;
  • one high-risk scenario with approval;
  • one denied-by-policy scenario;
  • sample JSONL traces in the docs.

Why it matters: a runnable reference package should not only show a happy path. It should teach production-like behavior.

5. A contribution kit for the community

Make external contribution easier:

  • a page called How to contribute patterns;
  • a template for new case studies;
  • a template for glossary entries;
  • a template for new policy templates.

Why it matters: a good open handbook grows faster when contributors know exactly how to help.

6. Stronger internal linking and chapter journeys

The book should make chapter movement easier:

  • clearer "what to read next" cues;
  • stronger links between architecture, case studies, and templates;
  • short decision paths inside parts.

Why it matters: the book becomes more useful when readers do not get lost between strong but dense sections.

7. A discoverability layer

Add a more systematic discoverability layer:

  • glossary;
  • cheatsheets;
  • stronger internal linking between chapters;
  • social preview assets;
  • more structured landing copy in ru/en/zh.

Why it matters: even a strong book does not help the community if people cannot find it, understand it quickly, and share it easily.

8. Social and sharing assets

The project should include lightweight sharing assets:

  • social preview assets;
  • a few shareable cheat sheets;
  • short landing summaries in ru/en/zh.

Why it matters: international reach grows when the book is easy not only to read, but also to share.

What to Do First

If only three steps happen next, I would do:

  1. Trace schema and event catalog
  2. Contribution kit
  3. Expand the case studies set

That would produce the fastest growth in practical value without rewriting the architectural chapters.

What Would Change in a Month

If this backlog is executed, the project will gain:

  • a much clearer entry point for new readers;
  • more pages people can quote and share;
  • more reusable artifacts for teams;
  • a better contribution path for the community;
  • a stronger international profile.

The Next Practical Step

If we follow this roadmap, the next best step is a trace schema and event catalog, followed by a contribution kit.