mise.toml

[settings]
experimental = true

[hooks]
postinstall = "if [ -z \"$CI\" ]; then lefthook install --force; fi"

[env]
# GDK root directory. Override with an absolute path, e.g: export GDK_ROOT=$HOME/workspace/gdk
# GDK_DIR is also accepted as an alias for GDK_ROOT.
# Do not use ~ (tilde is not expanded in variable values).
GDK_ROOT_RESOLVED = "{{env.GDK_ROOT | default(value=env.GDK_DIR | default(value=env.HOME ~ '/gitlab/gdk'))}}"


# Path to Siphon repository for proto compilation (siphon-proto crate)
# Clone from: https://gitlab.com/gitlab-org/analytics-section/siphon
SIPHON_PROTO_ROOT = "{{env.GDK_ROOT | default(value=env.GDK_DIR | default(value=env.HOME ~ '/gitlab/gdk'))}}/siphon"

# Auto-download prebuilt DuckDB shared library on first build (cached in target/).
DUCKDB_DOWNLOAD_LIB = "{{env.DUCKDB_DOWNLOAD_LIB | default(value='1')}}"

[tools]
"rust" = { version = "1.95.0", components = "rustfmt,clippy", targets = "aarch64-unknown-linux-gnu,x86_64-unknown-linux-gnu" }
"cargo:cargo-nextest" = "0.9.123"
"cargo:cargo-edit" = "0.13.8"
"cargo:cargo-audit" = "0.22.0"
"cargo:cargo-deny" = "0.19.0"
"cargo:cargo-geiger" = "0.13.0"
"cargo:cargo-cyclonedx" = "0.5.7"
"cargo:cargo-zigbuild" = "0.21.5"
"cargo:ast-grep" = "0.42.0"
"cargo:https://gitlab.com/gitlab-org/rust/gitlab-xtasks" = "tag:v1.5.0"
"cargo:cargo-bolero" = "0.13.4"
"sccache" = "0.13.0"
"zig" = "0.15.2"
clickhouse = "25.12"
duckdb = "1.5.1"
uv = "0.9.28"
node = "24"
markdownlint-cli2 = "0.22.1"
lychee = "0.23"
vale = "3.14.1"
lefthook = "2.1.3"
protoc = "34.1"
jq = "1.7"
yq = "4.52.5"
hyperfine = "1.20.0"
"aqua:gitleaks/gitleaks" = "8.30.1"
"github:google/go-jsonnet" = "0.22.0"

[tasks."hooks:install"]
run = ["lefthook install", "cp scripts/hooks/post-checkout $(git rev-parse --git-common-dir)/hooks/post-checkout"]
description = "Install git hooks (lefthook + worktree support)"

[tasks."fuzz:compile"]
run = "cargo +nightly bolero test fuzz_compile -p gkg-fuzz"
description = "Fuzz the query compiler (compile JSON → SQL pipeline)"

[tasks."fuzz:compile-structured"]
run = "cargo +nightly bolero test fuzz_compile_structured -p gkg-fuzz"
description = "Fuzz the query compiler with structured JSON query generation"

[tasks."fuzz:ruby"]
run = "cargo +nightly bolero test fuzz_ruby -p gkg-fuzz"
description = "Fuzz the Ruby code parser"

[tasks."fuzz:python"]
run = "cargo +nightly bolero test fuzz_python -p gkg-fuzz"
description = "Fuzz the Python code parser"

[tasks."fuzz:typescript"]
run = "cargo +nightly bolero test fuzz_typescript -p gkg-fuzz"
description = "Fuzz the TypeScript/JavaScript code parser"

[tasks."fuzz:java"]
run = "cargo +nightly bolero test fuzz_java -p gkg-fuzz"
description = "Fuzz the Java code parser"

[tasks."fuzz:kotlin"]
run = "cargo +nightly bolero test fuzz_kotlin -p gkg-fuzz"
description = "Fuzz the Kotlin code parser"

[tasks."fuzz:csharp"]
run = "cargo +nightly bolero test fuzz_csharp -p gkg-fuzz"
description = "Fuzz the C# code parser"

[tasks."fuzz:rust-parser"]
run = "cargo +nightly bolero test fuzz_rust_parser -p gkg-fuzz"
description = "Fuzz the Rust code parser"

[tasks."fuzz:indexer-messages"]
run = "cargo +nightly bolero test fuzz_indexer_messages -p gkg-fuzz"
description = "Fuzz indexer message deserialization"

[tasks.build]
run = "cargo build --release"
description = "Build the crates"

[tasks."build:cli"]
run = "cargo build --bin orbit --release --features duckdb-client/bundled"
description = "Build the CLI"

[tasks.test]
run = "./scripts/run-unit-tests.sh"
description = "Run unit + non-Docker integration tests (auto-discovers test targets)"

[tasks."test:fast"]
run = "CARGO_PROFILE_TEST_DEBUG=0 CARGO_PROFILE_TEST_SPLIT_DEBUGINFO=unpacked ./scripts/run-unit-tests.sh"
description = "Run unit + non-Docker integration tests with reduced debug info"

[tasks."test:local"]
run = "cargo nextest run --test local"
description = "Run local integration tests — compiler + querying pipeline (no Docker)"

[tasks."test:integration:codegraph"]
run = "cargo nextest run -p integration-tests-codegraph --no-capture"
description = "Run code-graph v2 validator tests — YAML Cypher suites (no Docker)"

[tasks."test:cli"]
run = "cargo build -p orbit-local && cargo nextest run --test cli --no-fail-fast"
description = "Run CLI integration tests — concurrency, worktrees (no Docker)"

[tasks."test:integration"]
run = "cargo nextest run --all-features --test containers --no-fail-fast --retries 0"
env = { DOCKER_HOST = "unix://{{env.HOME}}/.colima/gkg/docker.sock" }
description = "Run integration tests (requires Docker and `mise colima:start`)"

[tasks."test:integration:server"]
run = "cargo nextest run --all-features --test containers -j 4 --no-fail-fast --retries 0 -E 'test(data_correctness) | test(hydration_integration) | test(redaction_integration) | test(graph_formatter)'"
env = { DOCKER_HOST = "unix://{{env.HOME}}/.colima/gkg/docker.sock" }
description = "Run server integration tests (data correctness, hydration, redaction, graph formatter)"

[tasks."test:integration:data"]
run = "cargo nextest run --all-features --test containers --no-fail-fast --retries 0 --no-capture -E 'test(data_correctness)'"
env = { DOCKER_HOST = "unix://{{env.HOME}}/.colima/gkg/docker.sock" }
description = "Run data correctness integration tests"

[tasks."test:integration:indexer"]
run = "cargo nextest run --all-features --test containers --no-fail-fast --retries 0 -E 'test(indexer::)'"
env = { DOCKER_HOST = "unix://{{env.HOME}}/.colima/gkg/docker.sock" }
description = "Run indexer integration tests (schema migration, engine, NATS, namespace deletion)"

[tasks."test:integration:indexer:code"]
run = "cargo nextest run -p integration-tests --retries 0 -E 'test(code_indexing_task)'"
env = { DOCKER_HOST = "unix://{{env.HOME}}/.colima/gkg/docker.sock" }
description = "Run indexer integration tests (code indexing)"

[tasks.clean]
run = ["cargo clean", "rm -rf lib/*/"]
description = "Clean the crates"

[tasks."lint:fmt"]
run = "cargo fmt --all -- --check"
description = "Check Rust formatting"

[tasks."lint:code"]
run = [
  "cargo clippy --workspace --exclude duckdb-client --all-targets --all-features -- -D warnings",
  "cargo clippy -p duckdb-client --all-targets -- -D warnings",
]
description = "Run clippy lint checks"

[tasks."lint:code:fix"]
run = [
  "cargo fmt",
  "cargo clippy --workspace --exclude duckdb-client --all-targets --all-features --fix --allow-dirty -- -D warnings",
  "cargo clippy -p duckdb-client --all-targets --fix --allow-dirty -- -D warnings",
  "cargo fmt",
]
description = "Run clippy lint checks and fix applicable issues"

[tasks."fix:newlines"]
run = "gitlab-xtasks lint verify-newlines --file-extensions rs,md,yml,yaml,toml,astro,js,ts,json,mdx,vue,rb,css,mjs --directory . --fix"
description = "Fix newlines formatting in the codebase"

[tasks."local:index:repo"]
run = 'cargo run --release --bin orbit -- index ${usage_repo_path?} --stats'
description = "Index a local repository"
usage = 'arg "<repo_path>"'

[tasks."diff:branch"]
run = "git diff --unified=0 origin/${usage_base:-main}...HEAD $(xargs < .gitdiffignore) > git_diff_full.txt"
description = "Diff current branch against an origin branch (default: main)"
usage = 'arg "[base]" default="main" help="Branch on origin to diff against"'


[tasks."query:profile"]
run = 'cargo run -p query-profiler -- "$@"'
description = "Profile a GKG query against ClickHouse (use -- to pass args)"

[tasks."query:diff"]
run = 'python3 scripts/devtools/diff-profiler-results.py "$@"'
description = "Compare query-profiler result files (use -- to pass args)"

[tasks."perf:gtime"]
run = "./scripts/devtools/gtime.sh"

[tasks."perf:rss"]
run = './scripts/devtools/average_rss.sh ${usage_repo_path?} ${usage_interval:-0.025}'
description = "Measure average RSS memory usage during indexing"
usage = '''
arg "<repo_path>"
arg "[interval]" default="0.025" help="Sampling interval in seconds"
'''

[tasks."bench:download-repos"]
run = "gitlab-xtasks benchmark download-repos --config code-indexing-benchmark.yaml"
description = "Download benchmark repositories"

[tasks."bench:run"]
depends = ["bench:download-repos"]
run = "cargo build --bin orbit --release --features duckdb-client/bundled && gitlab-xtasks benchmark hyperfine --config code-indexing-benchmark.yaml --architecture local"
description = "Run code indexing benchmark"

[tasks."bench:v2"]
depends = ["bench:download-repos"]
run = """
cargo build --bin orbit --release --features duckdb-client/bundled
for repo in output/repos/*/; do
  name=$(basename "$repo")
  echo "=== Indexing $name with v2 ==="
  ./target/release/orbit index "$repo" --stats 2>&1 || echo "FAILED: $name"
done
"""
description = "Run v2 pipeline against all downloaded benchmark repos"

[tasks."bench:list-repos"]
run = "gitlab-xtasks benchmark list-repos --config code-indexing-benchmark.yaml"
description = "List benchmark repositories"

[tasks."bench:list-groups"]
run = "gitlab-xtasks benchmark list-groups --config code-indexing-benchmark.yaml"
description = "List benchmark repo groups"

[tasks."proto:go"]
run = "scripts/proto-go.sh generate"
description = "Generate Go protobuf/gRPC stubs for Workhorse integration"

[tasks."proto:go:check"]
run = "scripts/proto-go.sh check"
description = "Verify Go proto stubs are up to date"

[tasks."proto:ruby"]
run = '''
mkdir -p crates/gkg-server/proto/ruby
protoc --ruby_out=crates/gkg-server/proto/ruby \
  --grpc_out=crates/gkg-server/proto/ruby \
  --plugin=protoc-gen-grpc=$(which grpc_ruby_plugin) \
  -I crates/gkg-server/proto \
  crates/gkg-server/proto/gkg.proto
echo "Ruby proto files generated in crates/gkg-server/proto/ruby/"
'''
description = "Generate Ruby protobuf files for Rails client"

[tasks."proto:build-gem"]
run = '''
BUNDLE_GEMFILE=tools/protogem/Gemfile bundle install --quiet
ruby tools/protogem/build-proto-gem
'''
description = "Build gitlab-gkg-proto Ruby gem from proto definitions"

[tasks."proto:publish-gem"]
depends = ["proto:build-gem"]
run = 'gem push _build/gitlab-gkg-proto-*.gem'
description = "Publish gitlab-gkg-proto gem to RubyGems.org"

[tasks."server:start"]
run = 'GKG_GITLAB__JWT__VERIFYING_KEY=${GKG_GITLAB__JWT__VERIFYING_KEY:-$(cat $GDK_ROOT_RESOLVED/gitlab/.gitlab_knowledge_graph_secret 2>/dev/null || cat $GDK_ROOT_RESOLVED/gitlab/.gitlab_shell_secret 2>/dev/null || echo "development-secret-at-least-32-bytes")} cargo run -p gkg-server'
description = "Start the GKG server (webserver mode)"

[tasks."server:dispatch"]
run = './scripts/run-dispatcher.sh'
description = "Start the GKG server (dispatch-indexing mode)"

[tasks.dev]
run = [
  "mise run dev:check",
  "mise run dev:web ::: dev:indexer ::: dev:dispatcher",
]
description = "Start a local GKG environment (webserver + indexer + dispatcher) connected to GDK services"

[tasks.gdk]
run = "mise run dev"
description = "Alias for the local GKG environment connected to GDK services"

[tasks."dev:check"]
run = "./scripts/gkg-native-dev.sh check"
description = "Verify GDK-connected local development prerequisites"

[tasks."dev:status"]
run = "./scripts/gkg-native-dev.sh env"
description = "Show the derived GDK-backed environment for local development"

[tasks."dev:restart"]
run = "mise run dev"
description = "Restart the local GKG environment"

[tasks."dev:stop"]
run = "echo 'Use Ctrl+C in the running mise session to stop all dev processes'"
description = "Explain how to stop the local GKG environment"

[tasks."dev:setup"]
run = "./scripts/gkg-native-dev.sh setup"
description = "Create the graph database and apply config/graph.sql to ClickHouse"

[tasks."dev:web"]
run = "GKG_BIND_ADDRESS=127.0.0.1:${GKG_SERVER_PORT:-8090} GKG_GRPC_BIND_ADDRESS=127.0.0.1:${GKG_SERVER_GRPC_PORT:-50054} GKG_METRICS__PROMETHEUS__ENABLED=${GKG_ENABLE_METRICS:-false} GKG_METRICS__PROMETHEUS__PORT=${GKG_METRICS_PORT:-9100} ./scripts/gkg-native-dev.sh webserver"
description = "Start a local GKG webserver connected to GDK services"

[tasks."dev:indexer"]
run = "GKG_INDEXER_HEALTH_BIND_ADDRESS=127.0.0.1:${GKG_INDEXER_PORT:-4202} GKG_NATS__CONSUMER_NAME=${GKG_INDEXER_CONSUMER:-gkg-indexer-dev} GKG_METRICS__PROMETHEUS__ENABLED=${GKG_ENABLE_METRICS:-false} GKG_METRICS__PROMETHEUS__PORT=${GKG_INDEXER_METRICS_PORT:-9200} ./scripts/gkg-native-dev.sh indexer"
description = "Start a local GKG indexer connected to GDK services"

[tasks."dev:dispatcher"]
run = "./scripts/gkg-native-dev.sh dispatcher"
description = "Start a local GKG dispatch-indexing process connected to GDK services"

[tasks."dev:healthcheck"]
run = "./scripts/gkg-native-dev.sh healthcheck"
description = "Start the local GKG health-check process"

[tasks."dev:env"]
run = "./scripts/gkg-native-dev.sh env"
description = "Print the derived local GKG environment"


[tasks."ontology:validate"]
run = "find config/ontology -name '*.yaml' ! -name 'reference.yaml' | xargs uvx check-jsonschema --verbose --schemafile config/schemas/ontology.schema.json"
description = "Validate ontology YAML files against JSON schema"

[tasks."schema:generate"]
run = "cargo xtask schema -o config/schemas/config.schema.json"
description = "Generate JSON Schema for the server configuration"

[tasks."metrics:catalog"]
run = "cargo xtask metrics-catalog"
description = "Regenerate crates/gkg-observability/orbit-dashboards/gkg-metrics.json"

[tasks."metrics:catalog:check"]
run = "cargo xtask metrics-catalog --check"
description = "Fail if the committed metrics catalog is out of date"

[tasks."dashboards"]
run = "cargo xtask dashboards"
description = "Regenerate dashboards/orbit/*.dashboard.json from jsonnet sources"

[tasks."dashboards:check"]
run = "cargo xtask dashboards --check"
description = "Fail if the committed dashboards are out of date"

[tasks."schema:generate:ddl"]
run = """
cargo run --bin orbit -- debug ddl 2>/dev/null > config/graph.sql.tmp && mv config/graph.sql.tmp config/graph.sql
cargo run --bin orbit -- debug ddl-local 2>/dev/null > config/graph_local.sql.tmp && mv config/graph_local.sql.tmp config/graph_local.sql
"""
description = "Regenerate config/graph.sql (ClickHouse) and config/graph_local.sql (DuckDB) from ontology"

[tasks."lint:markdown"]
run = "markdownlint-cli2 README.md 'docs/**/*.md'"
description = "Run markdownlint on all markdown files"

[tasks."lint:vale"]
run = "vale --minAlertLevel error README.md docs"
description = "Run Vale prose linting on markdown files"

[tasks."lint:links"]
run = "lychee --offline --include-fragments --exclude-path output --exclude-path .claude --exclude-path .opencode '**/*.md'"
description = "Check for broken links in markdown files"

[tasks."lint:docs"]
depends = ["lint:markdown", "lint:vale", "lint:links"]
run = "echo 'All doc linters passed'"
description = "Run all documentation linters"

[tasks."skill:sync:orbit"]
run = "cp docs/source/remote/queries/query-language.md skills/orbit/references/query_language.md"
description = "Sync upstream docs into the Orbit agent skill (skills/orbit/references/)"