kami
597a26d551
Closes the COMPLETED-lie where a freestyle stage passes by producing a schema-valid artifact regardless of whether its build actually succeeded: a verify stage could run 'npm run build', watch it fail, then self-attest a build_verified artifact and transition to done. Tool failures never failed the stage. The compiler now flags the terminal stage autoBuildGate=true when no stage declares an explicit build_expectation. Code-ness is decided at runtime, not compile time (the declared kind is always the generic file_written; only the written paths reveal code): runExecutionGate folds the FileWritten manifest via sessionProducedBuildTarget() and promotes to a PROJECT build when a build manifest (package_json/gradle_module) or an imports_resolve-bearing path was written, then runs the real command from .correx/project.toml [commands]. A non-clean exit fails the stage retryably; a docs-only plan is left untouched. Live-proven (run 3, session 28812b44): gate fired on the real 'npm --prefix frontend run build', exit 2 on a bad tsconfig, RetryAttempted instead of a false COMPLETED. Modules green: transitions+workflow 67, kernel 60.
23 lines
1.8 KiB
TOML
23 lines
1.8 KiB
TOML
about = "CORREX — local-first, event-sourced orchestration kernel for LLM workflows. Kotlin 2.0 / JVM 21, Gradle multi-module, kotlinx.serialization, Ktor."
|
|
|
|
conventions = [
|
|
"Event log is the only source of truth; projections are disposable and rebuilt from events",
|
|
"Dependency direction: apps -> core -> infrastructure; no cross-core imports",
|
|
"No bare try-catch; use runCatching and sealed domain error types",
|
|
"Every new EventPayload must be registered in the eventModule polymorphic block (Serialization.kt)",
|
|
"Track multi-session or handoff work as native tasks: search before creating to avoid duplicates (task_search), task_context before starting, claim before working, submit_for_review when ready, complete after review. Don't open a task for a single self-contained edit you finish now. When a goal has dependency seams or independent review points, task_decompose it into a parent + DEPENDS_ON-linked children (one approval) instead of one big task; a session works one task at a time, so siblings are claimed by later runs as they unblock.",
|
|
]
|
|
|
|
[commands]
|
|
test-all = "./gradlew check --rerun-tasks"
|
|
test-module = "./gradlew :core:<module>:test --rerun-tasks"
|
|
context-lookup = "python scripts/ctx.py <query>"
|
|
epic-status = "bash scripts/epic-status.sh"
|
|
# Build-gate aliases (BuildExpectation → commandAlias): MODULE→typecheck, PROJECT→build, TESTS→test.
|
|
# The gate runner is whitespace-split / no-shell and runs at workspace_root (the repo), so `--prefix
|
|
# frontend` targets the Node subproject without a `cd`. `npm run build` = `tsc && vite build` (npm's
|
|
# own shell handles the &&), which fails on a dangling asset import — exactly the COMPLETED-lie hole.
|
|
typecheck = "npm --prefix frontend run build"
|
|
build = "npm --prefix frontend run build"
|
|
test = "npm --prefix frontend test"
|