Tooling

Leia treats tools as part of the language product.

Daily Loop

go run ./cmd/leia eval 'print(1 + 2 + 3)'
go run ./cmd/leia fmt --check tests/smoke/01_basic.leia
go run ./cmd/leia fmt --check --json tests/smoke/01_basic.leia
go run ./cmd/leia lint tests/smoke/01_basic.leia
go run ./cmd/leia lint --json tests/smoke/01_basic.leia
go run ./cmd/leia test tests/smoke/01_basic.leia
go run ./cmd/leia examples check examples/hello/dialects.leia
go run ./cmd/leia test --json --output test-report.json tests/smoke/01_basic.leia
go run ./cmd/leia check --no-docs --no-editor --no-examples .
go run ./cmd/leia check --quick .

leia check runs formatter, linter, .leia tests, manifest coverage, and docs checks, editor asset checks, and runnable repository example checks. Use skip flags only while iterating:

go run ./cmd/leia check --no-test --no-docs tests/smoke/01_basic.leia
go run ./cmd/leia check --json .

Use JSON reports when another tool or CI step needs stable machine-readable results:

go run ./cmd/leia capabilities --json
go run ./cmd/leia test --json --output test-report.json tests/smoke/01_basic.leia
go run ./cmd/leia evaluate --json --report eval-report.json examples/evaluate/basic_assert.leia

capabilities --json includes tooling.report_count and tooling.reports, the registry of CLI and release-script JSON reports. Each registry entry advertises status_field, scalar_fields, count_fields, collection_fields, and collection_item_fields when the report exposes machine-readable release evidence. Field names use dotted JSON paths; [] marks per-item array paths such as commands[].args. Entries with template: true contain placeholders such as FILE, DIST_DIR, or BIN_DIR; callers must substitute the listed template_parameters before executing the command. Release dashboards should read those fields instead of scraping human output; for example, scripts/run.sh public-blockers --json exposes blocker kind counts and scripts/run.sh production --list --json exposes runnable, skipped, and release-critical run/skip counts plus structured skip details and the configured release-critical names.

CI Profiles

go run ./cmd/leia ci smoke --list
go run ./cmd/leia ci smoke
go run ./cmd/leia ci pr --no-luajit
go run ./cmd/leia ci perf --no-luajit

Profiles:

Use --list before running a profile when you want to see exactly which shell commands will execute.

The checked-in CI profile is the source of truth for hosted or local automation. If hosted CI is unavailable, contributors can still reproduce the same command set locally:

go run ./cmd/leia ci smoke --list
go run ./cmd/leia ci pr --list
go run ./cmd/leia ci release --list
go run ./cmd/leia ci release --release-version vX.Y.Z --list
go run ./cmd/leia ci release --release-version vX.Y.Z --list --json
scripts/run.sh worktree --json

scripts/run.sh worktree --json reports finding_status_count and finding_statuses for dirty, prunable, missing, unreadable, and ahead/behind worktree summaries.

Run smoke before small changes, pr before review, and release only when preparing a tag or release candidate.

Manifest Checks

scripts/run.sh manifest-check tests benchmarks
scripts/run.sh worktree

The manifest check keeps test and benchmark discovery explicit so local and hosted validation run the same discovered coverage.

Modules

go run ./cmd/leia mod verify --json examples/ui/package_managed
go run ./cmd/leia mod verify --json examples/tooling/package_manager_workflow
go run ./cmd/leia mod graph --json examples/ui/package_managed
go run ./cmd/leia mod capability --json examples/ui/package_managed

leia mod manages local module metadata, require graphs, vendored dependency plans, lockfiles, native go.mod interop, and declared capability evidence. Use the modules reference for the full command surface.

Documentation

go run ./cmd/leia doc generate --layout site --output docs
go run ./cmd/leia doc generate --format json
go run ./cmd/leia doc check
go run ./cmd/leia doc check --json
scripts/run.sh docs

Generated reference pages are checked in. scripts/run.sh docs verifies generated CLI/stdlib/dialect references, spec HTML freshness, spec runnable examples, Markdown links, release reference coverage, retired naming, and documented repository script entrypoints. Use --json for machine-readable documentation evidence with failure_kinds and failure_details.

GitHub Pages publishes docs/ through .github/workflows/pages.yml. The workflow runs scripts/run.sh docs before building the site and scripts/run.sh site --site-dir ./_site after rendering.

Editors

scripts/run.sh editor
scripts/run.sh editor --require-tree-sitter
go run ./cmd/leia editor smoke

The editor check validates shared TextMate grammars, VS Code syntax assets, snippets, language configuration, extension JavaScript syntax, the spec preview helper, editor smoke fixtures, and tree-sitter corpus tests when the CLI is available. Use --require-tree-sitter in environments where the tree-sitter dependency is expected to be installed.

Examples

go run ./cmd/leia examples list
go run ./cmd/leia examples show repo-hello-fib
go run ./cmd/leia examples check examples/hello/fib.leia examples/hello/types_demo.leia examples/hello/dialects.leia
go run ./cmd/leia examples run repo-hello-fib

leia examples check is the same checker used by the repository-wide leia check examples step. Runnable examples execute locally; manual examples are reported as skipped with their required host capability, service, credential, or step-budget reason. When an example lives inside a directory with leia.mod, the checker validates that module with leia mod verify --json before running the example so package metadata cannot drift away from runnable project examples.

Playground

go run ./cmd/leia playground --help
go run ./cmd/leia playground --addr 127.0.0.1:8080

The browser playground serves local execution APIs plus runnable Playground, Tour, Examples, Evaluate, and AI tabs. It is backed by checked-in repository examples, so playground content is exercised by the same example and release validation as command-line examples.

Performance

go run ./cmd/leia bench --quick
go run ./cmd/leia bench compare --bench numeric/mandelbrot --runs 3 --warmup 1
go run ./cmd/leia bench strict --bench table/table_array_access --runs 3 --warmup 1 \
  --json /tmp/leia-strict.json \
  --markdown /tmp/leia-strict.md
scripts/run.sh perf --feature-smoke
scripts/run.sh perf --validate-only /tmp/leia_performance_gate/timing_gate.json --json

Use --no-luajit when LuaJIT is not installed or when a benchmark has no useful Lua reference. When LuaJIT data is present and script-timed, the check also enforces --luajit-threshold so published performance claims remain tied to measured validation instead of report-only comparisons. See the performance reference.

Diagnostics

go run ./cmd/leia diag bundle --output /tmp/leia-diag --skip-benchmarks
go run ./cmd/leia diag bundle --output /tmp/leia-diag --skip-go-tests --skip-benchmarks --json
scripts/run.sh diag table/table_array_access

Use diagnostics bundles when filing performance or correctness issues. They collect environment, docs/test status, and optional benchmark summaries.

Release Evidence

go run ./cmd/leia capabilities --json
scripts/run.sh production --quick
scripts/run.sh production --quick --list --json
scripts/run.sh production --quick --list --out-dir /tmp/leia-release-plan
scripts/run.sh test release-matrix
scripts/run.sh editor --json
scripts/run.sh release-dist --json
scripts/run.sh release-check --json --version vX.Y.Z
scripts/run.sh release-check --build

The release process is documented in Release Process.