Taskfile.yml

version: "3"

tasks:
  setup:
    desc: "Set up the development environment"
    cmds:
      - go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.14.0 || true
      - task: governance:hooks:install
      - go mod download

  default:
    desc: Show available tasks
    cmds:
      - task --list

  update:
    desc: "Update Go modules and the docs lockfile"
    cmds:
      - go get -u all
      - go mod tidy
      - uv lock --upgrade

  upgrade:
    desc: "Upgrade all dependencies to latest then update"
    cmds:
      - task: update

  set-version:
    desc: "Set version across all packages (usage: task set-version -- 1.2.3)"
    requires:
      vars: [CLI_ARGS]
    vars:
      VERSION: "{{.CLI_ARGS}}"
    cmds:
      - |
        set -euo pipefail

        echo "Setting version to {{.VERSION}} across all packages..."

        # Go CLI version
        sed -i '' 's/Version  = ".*"/Version  = "{{.VERSION}}"/' cmd/commands/root.go
        sed -i '' 's/var version = ".*"/var version = "{{.VERSION}}"/' cmd/ai-rulez/main.go

        # npm release wrapper
        sed -i '' 's/"version": ".*"/"version": "{{.VERSION}}"/' release/npm/package.json

        # pyproject.toml
        sed -i '' 's/^version = ".*"/version = "{{.VERSION}}"/' pyproject.toml

        # PyPI wrapper __version__
        sed -i '' 's/__version__ = ".*"/__version__ = "{{.VERSION}}"/' release/pypi/__init__.py
        sed -i '' 's/__version__ = ".*"/__version__ = "{{.VERSION}}"/' release/pypi/ai_rulez/__init__.py

        # Pre-commit hook rev (internal/hooks/setup.go)
        sed -i '' 's/officialPreCommitRev  = "v[^"]*"/officialPreCommitRev  = "v{{.VERSION}}"/' internal/hooks/setup.go

        # Pre-commit bootstrap script default version (scripts/pre-commit/run-ai-rulez.sh)
        sed -i '' 's/AI_RULEZ_VERSION="\${AI_RULEZ_VERSION:-v[^"]*}"/AI_RULEZ_VERSION="\${AI_RULEZ_VERSION:-v{{.VERSION}}}"/' scripts/pre-commit/run-ai-rulez.sh

        # README pre-commit example + poly hook source revision
        sed -i '' 's/rev: v[0-9][0-9.]*/rev: v{{.VERSION}}/' README.md
        sed -i '' 's/revision = "v[0-9][0-9.]*"/revision = "v{{.VERSION}}"/g' README.md

        # Poly hook catalog + docs: pinned ai-rulez versions (npx/uvx) and revisions
        for f in poly-hooks.toml docs/poly-hooks.md; do
          sed -i '' 's/ai-rulez@[0-9][0-9.]*/ai-rulez@{{.VERSION}}/g' "$f"
          sed -i '' 's/ai-rulez==[0-9][0-9.]*/ai-rulez=={{.VERSION}}/g' "$f"
        done
        sed -i '' 's/revision = "v[0-9][0-9.]*"/revision = "v{{.VERSION}}"/g' docs/poly-hooks.md

        echo "Done. Updated version to {{.VERSION}} in:"
        echo "  - cmd/commands/root.go"
        echo "  - cmd/ai-rulez/main.go"
        echo "  - release/npm/package.json"
        echo "  - pyproject.toml"
        echo "  - release/pypi/__init__.py"
        echo "  - release/pypi/ai_rulez/__init__.py"
        echo "  - internal/hooks/setup.go"
        echo "  - scripts/pre-commit/run-ai-rulez.sh"
        echo "  - README.md"
        echo "  - poly-hooks.toml"
        echo "  - docs/poly-hooks.md"

  setup:llm:
    desc: "Download, sha256-verify and cache the static liter-llm library; prints the CGO_LDFLAGS export (needs the network once)"
    cmds:
      - bash scripts/setup-liter-llm.sh

  build:
    desc: "Build the CLI binary as a release does: liter-llm linked statically (plain `go build` links the shared library the module ships, which only runs on this machine)"
    cmds:
      - bash -c 'if [ -z "${CGO_LDFLAGS:-}" ]; then eval "$(bash scripts/setup-liter-llm.sh)"; fi; go build -ldflags="-s -w" -o bin/ai-rulez ./cmd/ai-rulez'

  test:release:
    desc: "Run the model-layer tests against the static liter-llm library, the way CI does"
    cmds:
      - bash -c 'if [ -z "${CGO_LDFLAGS:-}" ]; then eval "$(bash scripts/setup-liter-llm.sh)"; fi; go test ./internal/llm/... ./internal/pricing/... ./internal/tokens/... ./cmd/...'

  release:dry-run:
    desc: "Build this platform's release archive locally and check what the wrappers expect (see docs/maintainers/release.md)"
    cmds:
      - bash scripts/release-dry-run.sh

  test:
    desc: "Run unit and integration tests"
    vars:
      TEST_TIMEOUT: '{{default "30m" .TEST_TIMEOUT}}'
      COMMAND_TEST_LOG_DIR: '{{default "" .COMMAND_TEST_LOG_DIR}}'
    cmds:
      - '{{if eq OS "windows"}}go run ./scripts/testshard -shards 4 -log-dir "{{.COMMAND_TEST_LOG_DIR}}"{{else}}go test{{end}} -timeout {{.TEST_TIMEOUT}} ./cmd/... ./internal/... ./pkg/... ./scripts/... ./bench/... ./schema/... ./tests/fixtures/... ./tests/integration/... ./tests/testutil/... ./tests/archlint/...'

  schema:local:generate:
    desc: "Derive the local overlay schema from the main schema"
    env:
      UPDATE_LOCAL_SCHEMA: "1"
    cmds:
      - go test ./schema -run TestLocalSchemaInSync -count=1

  test:gemini-settings:
    desc: "Verify Gemini extension settings schema and native projection"
    cmds:
      - go test ./internal/config ./internal/generator/plugin ./schema -count=1

  test:golden:update:
    desc: "Rewrite golden files (UPDATE_GOLDEN=1) in every package that has them, then the tokens golden; review the diff"
    env:
      UPDATE_GOLDEN: "1"
    cmds:
      - go test -count=1 ./cmd/commands/ ./internal/lint/ ./internal/evals/evalimport/ ./internal/importer/ ./internal/mcp/ ./internal/okfbridge/ ./internal/generator/... ./internal/publish/... ./internal/migrate/ ./internal/agentplugins/ ./internal/telemetry/ ./internal/ard/ ./internal/policy/ ./tests/golden/ ./tests/archlint/
      - go test -count=1 ./internal/generator -run Golden_TokenReportJSON -update-tokens-golden

  test:verbose:
    desc: "Run all tests with verbose output"
    cmds:
      - go test -v ./...

  test:race:
    desc: "Run tests with race detector"
    cmds:
      - go test -race ./...

  test:e2e:
    desc: "Run all e2e tests"
    cmds:
      - go test -timeout 30m ./tests/e2e/...

  test:e2e:cli:
    desc: "Run e2e CLI tests"
    cmds:
      - go test ./tests/e2e/cli/...

  test:e2e:mcp:
    desc: "Run e2e MCP tests"
    cmds:
      - go test ./tests/e2e/mcp/...

  test:e2e:integration:
    desc: "Run e2e integration tests"
    cmds:
      - go test ./tests/e2e/integration/...

  test:otelcol:
    desc: "Send telemetry to a pinned real OpenTelemetry Collector (needs docker)"
    env:
      AI_RULEZ_E2E_OTELCOL: "1"
    cmds:
      - go test ./internal/telemetry -run 'TestCollector' -count=1 -v

  test:platform:
    desc: "Run cross-platform specific tests"
    cmds:
      - go test -timeout 30m ./tests/platform/...

  test:bats:
    desc: "Run the Bats suites for shell scripts (usage: task test:bats -- [bats args or suite path])"
    cmds:
      # run.sh sets BATS_LIB_PATH=tests/lib (bats_load_library searches nothing else) and builds
      # ai-rulez once for the suites that render scripts through the binary.
      - scripts/bats/run.sh {{.CLI_ARGS}}

  test:lib:sync:
    desc: "Refresh tests/lib/xberg-bats from xberg-io/actions (task test:lib:sync -- --ref v1 re-pins)"
    cmds:
      - scripts/bats/sync-xberg-bats.sh {{.CLI_ARGS}}

  test:lib:check:
    desc: "Fail when tests/lib/xberg-bats has drifted from the pinned xberg-io/actions commit"
    cmds:
      - scripts/bats/sync-xberg-bats.sh --check

  lint:bats:
    desc: "Format-check and lint the Bats suites (poly has no engine for .bats)"
    cmds:
      - scripts/bats/lint.sh

  bench:judge:
    desc: "Run the judge benchmark on Amazon Bedrock: the synthetic corpus of bench/judge through the model pipeline, then precision/recall per code, the trap false-positive rate, grounding and cost. Spends money (hard cap 5 USD by default); pass flags after --, for example: task bench:judge -- --max-cost 3 --only mal-s03-a,trap-t01. Uses AWS_PROFILE=default and AWS_REGION=eu-north-1 unless --aws-profile / --aws-region say otherwise"
    cmds:
      - go run ./bench/judge/cmd/judgebench {{.CLI_ARGS}}

  bench:judge:offline:
    desc: "Run the offline half of the judge benchmark: the verifier and the convention attachment on the corpus, no model"
    cmds:
      - go test ./bench/judge/... -count=1

  test:benchmark:
    desc: "Run Go benchmarks (small and medium synthetic projects)"
    cmds:
      - go test -bench=. -run='^$' -timeout 30m ./...

  test:benchmark:large:
    desc: "Run Go benchmarks including the 2000-file synthetic project (one iteration each)"
    env:
      AI_RULEZ_BENCH_LARGE: "1"
    cmds:
      - go test -bench=. -benchtime=1x -run='^$' -timeout 60m ./...

  test:benchmark:regression:
    desc: "Run the curated benchmark subset and fail on a >25% time/op regression against scripts/bench-baseline.txt"
    cmds:
      - scripts/bench.sh

  test:benchmark:baseline:
    desc: "Re-record scripts/bench-baseline.txt on this machine"
    cmds:
      - scripts/bench.sh --update

  test:all:
    desc: "Run all test suites"
    cmds:
      - task: test
      - task: test:platform
      - task: test:e2e

  lint:
    desc: "Run poly formatting and lint checks"
    cmds:
      - poly fmt --check .
      - poly lint .

  check:
    desc: "Run go vet and golangci-lint"
    cmds:
      - go vet ./...
      - golangci-lint run --timeout=5m

  format:
    desc: "Format Go code"
    cmds:
      - go fmt ./...

  generate:
    desc: "Generate ai-rulez configuration files"
    cmds:
      - go run ./cmd/ai-rulez generate

  validate:
    desc: "Validate ai-rulez configuration"
    cmds:
      - go run ./cmd/ai-rulez validate

  docs:cli:
    desc: "Regenerate docs/cli-reference.md and the command index of docs/cli.md from the Cobra command tree"
    cmds:
      - go run ./scripts/clidocs

  docs:cli:check:
    desc: "Fail when docs/cli-reference.md or the command index of docs/cli.md differ from the command tree"
    cmds:
      - go run ./scripts/clidocs --check

  docs:llms:
    desc: "Regenerate docs/llms.txt and docs/llms-full.txt from zensical.toml and docs/"
    cmds:
      - go run ./scripts/docsllms

  docs:llms:check:
    desc: "Fail when docs/llms.txt or docs/llms-full.txt differ from the docs sources"
    cmds:
      - go run ./scripts/docsllms --check

  harness:verify:
    desc: "List harness trap and limit rows with source and verified_on; fail when older than max_table_age_days (default 90). HARNESS_MAX_AGE_DAYS overrides; HARNESS_VERIFY_FETCH=1 also re-fetches each quote (network)"
    cmds:
      - HARNESS_VERIFY=1 go test ./internal/lint -run TestHarnessVerify -count=1 -v

  demo:record:
    desc: "Re-record the README terminal demos with vhs (all, or name them: task demo:record -- validate); uses the ai-rulez on PATH"
    cmds:
      - bash scripts/demo/run.sh record {{.CLI_ARGS}}

  demo:check:
    desc: "Parse every demo tape and check each committed GIF exists and is at most 1 MiB"
    cmds:
      - bash scripts/demo/run.sh check

  clean:
    desc: "Clean build artifacts"
    cmds:
      - rm -rf bin/
      - go clean -cache -testcache

  governance:hooks:install:
    desc: Provision machine preferences and install reviewed AI governance hooks
    cmds:
      - |
        python3 - <<'PYTHON'
        import pathlib
        import shutil
        import subprocess
        import tomllib

        if shutil.which("ai-rulez") is None:
            raise SystemExit("Install the reviewed ai-rulez CLI before installing hooks")
        preferences = pathlib.Path("poly.local.toml")
        text = preferences.read_text() if preferences.exists() else ""
        data = tomllib.loads(text)
        if "hook_preferences" not in data:
            preferences.write_text(text.rstrip() + '\n\n[hook_preferences]\nchannels = ["system"]\n')
        elif "system" not in data["hook_preferences"].get("channels", []):
            raise SystemExit("Add system to hook_preferences.channels in poly.local.toml")
        subprocess.run(["poly", "hooks", "install"], check=True)
        PYTHON

  governance:check:
    desc: Check materialized local governance outputs against shared sources
    cmds:
      - ai-rulez generate --check --no-local --strict-config