SpineEventEngine · alexander-yevsyukov · Jul 23, 2026 · Jul 21, 2026 · Jul 22, 2026 · Jul 22, 2026
diff --git a/.gitmodules b/.gitmodules
@@ -10,9 +10,6 @@
 [submodule "tool-base"]
 	path = tool-base
 	url = https://github.com/SpineEventEngine/tool-base.git
-[submodule "delivery-server"]
-	path = delivery-server
-	url = https://github.com/SpineEventEngine/delivery-server.git
 [submodule "config"]
 	path = config
 	url = https://github.com/SpineEventEngine/config.git
@@ -49,9 +46,15 @@
 [submodule "money"]
 	path = money
 	url = https://github.com/SpineEventEngine/money.git
+[submodule "gcloud-jvm"]
+	path = gcloud-jvm
+	url = https://github.com/SpineEventEngine/gcloud-jvm.git
 [submodule ".agents/shared"]
 	path = .agents/shared
 	url = https://github.com/SpineEventEngine/agents.git
 	branch = master
 	update = merge
 	ignore = all
+[submodule "agents"]
+	path = agents
+	url = https://github.com/SpineEventEngine/agents.git
diff --git a/README.md b/README.md
@@ -1,2 +1,119 @@
 # summit
-The repository for automation of cross-repository works
+
+The meta-repository for automating cross-repository work across the [Spine SDK][spine-org].
+
+`summit` is a Git superproject: it assembles the Spine SDK repositories as submodules
+in a single working tree, so agents and the shared `config`/CI tooling can act across
+the whole SDK at once. It ships no code and has no build of its own — its content is the
+pinned commits of the submodules below, plus the shared agent tooling under `.agents/`.
+
+## Repository layout
+
+The submodules are pinned to fixed commits, giving a reproducible snapshot of the
+SDK. They are grouped below by function, not by strict dependency order.
+
+### Shared infrastructure
+
+| Submodule | Role |
+|-----------|------|
+| [`config`][config] | Dependencies and build configurations shared among subprojects |
+
+The shared agent tooling — skills, scripts, and guidelines — lives in the
+[`agents`][agents] repository, mounted at `.agents/shared`. Unlike the submodules
+listed here, that mount floats to the tip of `master` rather than a pinned commit.
+The same repository is also wired in as a pinned top-level submodule at `agents/`,
+so cross-repository automation can run against the tooling repo itself, just as it
+does against the SDK repositories above.
+
+### Foundation & utilities
+
+| Submodule | Role |
+|-----------|------|
+| [`base-libraries`][base-libraries] | The framework's base types and utilities |
+| [`reflect`][reflect] | Utilities for working with reflection in Java and Kotlin projects |
+| [`logging`][logging] | Fluent logging API for Kotlin projects |
+| [`testlib`][testlib] | Testing utilities for Spine SDK development and users |
+
+### Domain & value types
+
+| Submodule | Role |
+|-----------|------|
+| [`base-types`][base-types] | Popular value object types and associated code |
+| [`change`][change] | Data types and utilities for changes and mismatches in data values |
+| [`time`][time] | Protobuf-based date/time types and utilities |
+| [`money`][money] | Currency and money data types and operations |
+
+### Compiler & code generation
+
+| Submodule | Role |
+|-----------|------|
+| [`compiler`][compiler] | The Spine Compiler — extendable Protobuf compilation |
+| [`tool-base`][tool-base] | Common code for development tools |
+| [`ProtoTap`][ProtoTap] | Utilities for tapping `protoc` output |
+| [`validation`][validation] | Library and Compiler plugins for generating custom validation code |
+| [`core-jvm-compiler`][core-jvm-compiler] | Plugins of the CoreJvm library for the Spine Compiler |
+
+### Framework core
+
+| Submodule | Role |
+|-----------|------|
+| [`core-jvm`][core-jvm] | The JVM-based implementation of the Spine framework core |
+
+### Storage & runtime
+
+| Submodule | Role |
+|-----------|------|
+| [`jdbc-storage`][jdbc-storage] | Support for storage in JDBC-compliant databases |
+| [`gcloud-jvm`][gcloud-jvm] | Support for Spine-based Kotlin and Java apps on Google Cloud |
+
+## Getting started
+
+Clone with all submodules in one step:
+
+```bash
+git clone --recursive https://github.com/SpineEventEngine/summit.git
+```
+
+If you cloned without `--recursive`, or you are working in a fresh `git worktree` or
+a shallow checkout, the submodules start uninitialised and the `.agents` symlinks
+dangle. Bootstrap them with:
+
+```bash
+./init-submodules   # materialise the config-managed submodules at their pinned commits
+./config/pull       # float the shared submodules and copy in the shared files
+```
+
+`./init-submodules` runs automatically at the start of a Claude Code session. See
+[`AGENTS.md`](AGENTS.md) for the full bootstrap chain and the rationale behind it.
+
+## Working across repositories
+
+`summit` is where cross-repository work happens: coordinated agent sessions and the
+shared `config` tooling propagate changes — dependency bumps, CI workflows, agent
+guidelines — across the SDK from one place. Agent orientation, safety rules, and the
+available skills are described in [`AGENTS.md`](AGENTS.md).
+
+## License
+
+`summit`, like the rest of the Spine SDK, is distributed under the
+[Apache License 2.0](LICENSE).
+
+[spine-org]: https://github.com/SpineEventEngine
+[agents]: https://github.com/SpineEventEngine/agents
+[config]: https://github.com/SpineEventEngine/config
+[base-libraries]: https://github.com/SpineEventEngine/base-libraries
+[reflect]: https://github.com/SpineEventEngine/reflect
+[logging]: https://github.com/SpineEventEngine/logging
+[testlib]: https://github.com/SpineEventEngine/testlib
+[base-types]: https://github.com/SpineEventEngine/base-types
+[change]: https://github.com/SpineEventEngine/change
+[time]: https://github.com/SpineEventEngine/time
+[money]: https://github.com/SpineEventEngine/money
+[compiler]: https://github.com/SpineEventEngine/compiler
+[tool-base]: https://github.com/SpineEventEngine/tool-base
+[ProtoTap]: https://github.com/SpineEventEngine/ProtoTap
+[validation]: https://github.com/SpineEventEngine/validation
+[core-jvm-compiler]: https://github.com/SpineEventEngine/core-jvm-compiler
+[core-jvm]: https://github.com/SpineEventEngine/core-jvm
+[jdbc-storage]: https://github.com/SpineEventEngine/jdbc-storage
+[gcloud-jvm]: https://github.com/SpineEventEngine/gcloud-jvm
diff --git a/agents b/agents
diff --git a/base-libraries b/base-libraries
diff --git a/base-types b/base-types
diff --git a/delivery-server b/delivery-server
diff --git a/docs/project.md b/docs/project.md
@@ -1,18 +1,44 @@
-<!-- Template — copy to a sibling repo's .agents/project.md and fill in the
-     sections below. In the config repo itself this file is a template only. -->
-
-# Project: <name>
+# Project: summit
 
 ## Overview
 
-*One paragraph: what this repo is, what problem it solves, and its role in the
-Spine SDK organisation.*
+`summit` is a Git superproject (a meta-repository) that assembles the Spine SDK
+repositories as submodules in a single working tree. It exists to automate
+cross-repository work across the SDK: coordinated agent sessions and the shared
+`config`/CI machinery act across many repositories at once. `summit` ships no code and
+has no build of its own — its content is the pinned commits of the submodules it
+aggregates, plus the shared agent tooling under `.agents/`.
 
 ## Architecture
 
-*Role in the org: library / tool / Gradle plugin / application.
-Key patterns, public API boundaries, and constraints specific to this repo.*
+Role in the organisation: a **coordination superproject** — not a library, tool,
+Gradle plugin, or application.
+
+- **Submodules.** The SDK repositories listed in `.gitmodules` are pinned to fixed
+  commits, giving a reproducible snapshot of the whole SDK. The shared
+  `.agents/shared` submodule is the exception: it declares a tracked `branch` and
+  floats to the tip of `master`, so shared skills, scripts, and guidelines stay
+  current with no file churn in consumer pull requests.
+- **Bootstrapping.** A fresh `git worktree` or shallow checkout starts with the
+  submodules uninitialised, so the `.agents` symlinks dangle. Run `./init-submodules`
+  to materialise the config-managed submodules at their pinned commits, then
+  `./config/pull` to float the shared submodules and copy the shared files in. Claude
+  Code runs `./init-submodules` automatically via a `SessionStart` hook.
+- **Not a JVM build.** `summit` has no Gradle build, coding style, or tests of its
+  own, so the shared JVM requirements in `.agents/guidelines/jvm-project.md` do not
+  govern this repository. The aggregated repositories are JVM (Kotlin/Java) projects
+  and are each subject to that guideline in their own right.
+
+<!-- `summit` has no build of its own; the jvm-project.md link the template offers is
+     intentionally left commented out. -->
+
+## Cross-repository workflows
+
+Repeatable, parameterised procedures for acting across the SDK repositories live under
+[`docs/rollout/`](rollout/). Each pairs a repo-owned script (the deterministic
+mechanics) with an agent-driven playbook (the judgement steps):
 
-<!-- JVM projects: uncomment the line below after seeding this file.
-Read [`.agents/guidelines/jvm-project.md`](.agents/guidelines/jvm-project.md) for build stack, coding style, tests, and versioning.
--->
+- [`rollout/proofread.md`](rollout/proofread.md) — run the `proofread` skill across a
+  repo end-to-end (bump → build → sweep → pre-PR → PR), driven by the
+  [`proofread-repo`](../proofread-repo) script, with an optional
+  [`proofread-fanout.workflow.js`](rollout/proofread-fanout.workflow.js) for the sweep.