@frankzhang2026/opencode-android-orchestrator 1.0.5 → 1.2.0

package/templates/automation/tasks/TASK-TEMPLATE.json.example

@@ -1,5 +1,5 @@
 {
-  "schemaVersion": 3,
+  "schemaVersion": 4,
   "id": "TASK-EXAMPLE-001",
   "title": "Replace with one small, observable behavior change",
   "designApproved": true,
@@ -37,9 +37,7 @@
     "android-orchestrator-verification-before-completion"
   ],
   "acceptanceCriteria": [
-    "Replace with an observable input/output or UI behavior",
-    "A focused test demonstrates the behavior and is captured RED before implementation",
-    "Debug build passes; unit tests and Android lint also pass when their configuration flags are enabled"
+    "Replace with an observable input/output or UI behavior"
   ],
   "nonGoals": [
     "Do not refactor unrelated screens",
@@ -47,7 +45,7 @@
   ],
   "targetTests": [
     {
-      "gradleTask": "testDebugUnitTest",
+      "gradleTask": ":app:testDebugUnitTest",
       "filter": "*ReplaceWithFocusedTest"
     }
   ],
@@ -55,7 +53,7 @@
   "testPolicy": "required",
   "testPolicyReason": "Behavior changes require a focused regression test",
   "verification": {
-    "version": 1,
+    "version": 2,
     "maxPreparationFixes": 1,
     "cases": [
       {
@@ -72,9 +70,17 @@
         },
         "expectedFailure": {
           "type": "java.lang.AssertionError",
-          "origin": "The assertion at the approved behavior call"
+          "origin": {
+            "className": "ReplaceWithFocusedTest",
+            "methodName": "replace with observable behavior"
+          }
         }
       }
-    ]
+    ],
+    "inventory": {
+      "mode": "focusedBaseline",
+      "existingSkips": "reject",
+      "emptyBaseline": "reject"
+    }
   }
 }

package/templates/automation/verification/collect.init.gradle

@@ -0,0 +1,108 @@
+import groovy.json.JsonOutput
+import groovy.json.JsonSlurper
+import org.gradle.api.tasks.testing.Test
+import org.gradle.api.tasks.testing.TestDescriptor
+import org.gradle.api.tasks.testing.TestListener
+import org.gradle.api.tasks.testing.TestResult
+
+// All three phases use this collector. Never infer execution from existing XML.
+// Gradle 6.7.1 does not expose command-line -D properties through System yet
+// while evaluating init scripts in a (including single-use) daemon.
+def requestPath = gradle.startParameter.systemPropertiesArgs['orchestrator.inventoryRequest'] ?: System.getProperty('orchestrator.inventoryRequest')
+if (!(requestPath instanceof String) || requestPath.trim().isEmpty()) {
+    throw new GradleException('Missing orchestrator.inventoryRequest: pass an absolute request JSON path with -D')
+}
+def requestFile = new File(requestPath)
+if (!requestFile.isAbsolute() || !requestFile.isFile() || !requestFile.canRead()) {
+    throw new GradleException('Invalid orchestrator.inventoryRequest: an existing readable absolute file is required')
+}
+def request = new JsonSlurper().parse(requestFile)
+def outputFile = new File(request.output as String)
+def append = { Map value ->
+    synchronized (outputFile) {
+        outputFile << JsonOutput.toJson(value + [runId: request.runId]) << '\n'
+    }
+}
+append([kind: 'start', phase: request.phase, gradleVersion: gradle.gradleVersion])
+def filtersByTask = request.targets.groupBy { it.gradleTask }.collectEntries { path, targets ->
+    [(path): targets.collect { it.filter }.unique().sort()]
+}
+gradle.taskGraph.whenReady { graph ->
+    filtersByTask.each { path, filters ->
+        def task = graph.allTasks.find { it.path == path }
+        if (!(task instanceof Test)) throw new GradleException("Inventory target must resolve to an exact Test task: ${path}")
+        def roots = [] as Set
+        def classes = task.testClassesDirs.files.collect { it.canonicalFile } as Set
+        def sourceSets = task.project.extensions.findByName('sourceSets')
+        sourceSets?.each { sourceSet ->
+            if (sourceSet.output.classesDirs.files.any { classes.contains(it.canonicalFile) }) {
+                roots.addAll(sourceSet.allSource.srcDirs)
+                roots.addAll(sourceSet.resources.srcDirs)
+            }
+        }
+        // AGP variants expose the source providers actually used by each unit test.
+        def android = task.project.extensions.findByName('android')
+        if (android != null && android.hasProperty('unitTestVariants')) {
+            android.unitTestVariants.each { variant ->
+                if (task.name == "test${variant.name.capitalize()}") {
+                    variant.sourceSets.each { provider ->
+                        roots.addAll(provider.javaDirectories)
+                        roots.addAll(provider.resourcesDirectories)
+                        if (provider.hasProperty('kotlinDirectories')) roots.addAll(provider.kotlinDirectories)
+                    }
+                }
+            }
+        }
+        // Modern AGP also exposes variant-specific test source sets through its DSL.
+        // Only unit-test providers are classified here, never production providers.
+        if (android != null && android.hasProperty('sourceSets')) {
+            android.sourceSets.findAll { it.name == 'test' || it.name.startsWith('test') && it.name.length() > 4 && Character.isUpperCase(it.name.charAt(4)) }.each { sourceSet ->
+                roots.addAll(sourceSet.java.srcDirs)
+                roots.addAll(sourceSet.resources.srcDirs)
+                if (sourceSet.hasProperty('kotlin')) roots.addAll(sourceSet.kotlin.srcDirs)
+                roots.addAll(sourceSet.assets.srcDirs)
+                roots.addAll(sourceSet.res.srcDirs)
+            }
+        }
+        append([kind: 'task', taskPath: path, filters: filters,
+            sourceRoots: roots.collect { it.absoluteFile.toPath().normalize().toString() }.sort(),
+            testClassesDirs: classes.collect { it.toString() }.sort()])
+        task.outputs.upToDateWhen { false }
+        task.outputs.doNotCacheIf('Orchestrator inventory requires actual test execution') { true }
+        task.ignoreFailures = true
+        task.failFast = false
+        if (task.hasProperty('dryRun')) task.dryRun = false
+        // An approved new test class need not exist before Coder starts.
+        task.filter.failOnNoMatchingTests = request.phase != 'baseline'
+        if (task.hasProperty('failOnNoDiscoveredTests')) task.failOnNoDiscoveredTests = request.phase != 'baseline'
+        task.filter.setIncludePatterns(*(filters as String[]))
+        def reportDir = new File(outputFile.parentFile, 'reports/' + path.substring(1).replace(':', '/'))
+        task.reports.junitXml.required.set(true)
+        task.reports.junitXml.outputLocation.set(reportDir)
+        task.addTestListener(new TestListener() {
+            void beforeSuite(TestDescriptor descriptor) {}
+            void beforeTest(TestDescriptor descriptor) {}
+            void afterTest(TestDescriptor descriptor, TestResult result) {
+                def failures = result.exceptions.collect { failure ->
+                    [type: failure.class.name, message: failure.message ?: '', stack: failure.stackTrace.collect { it.toString() }]
+                }
+                append([kind: 'case', taskPath: path, className: descriptor.className ?: '',
+                    name: descriptor.name ?: '', displayName: descriptor.displayName,
+                    result: result.resultType.toString(), failures: failures])
+            }
+            void afterSuite(TestDescriptor descriptor, TestResult result) {
+                if (descriptor.parent == null) append([kind: 'suite', taskPath: path,
+                    tests: result.testCount, failures: result.failedTestCount, skipped: result.skippedTestCount])
+            }
+        })
+    }
+}
+gradle.taskGraph.afterTask { task, state ->
+    if (filtersByTask.containsKey(task.path)) append([kind: 'taskEnd', taskPath: task.path,
+        executed: state.executed, skipped: state.skipped, noSource: state.noSource,
+        upToDate: state.upToDate, skipMessage: state.skipMessage,
+        failure: state.failure == null ? null : state.failure.toString()])
+}
+gradle.buildFinished { result ->
+    append([kind: 'end', failure: result.failure == null ? null : result.failure.toString()])
+}

package/templates/automation/verification/contract.cjs

@@ -0,0 +1,119 @@
+"use strict";
+
+// One V4/V5 validator is used by both the queue and the installed shell entry.
+const assert = (condition, message) => { if (!condition) throw new Error(message); };
+const object = value => value !== null && typeof value === "object" && !Array.isArray(value);
+const text = value => typeof value === "string" && value.length > 0 && !/[\u0000\r\n]/.test(value);
+const fields = (value, required, optional = []) => object(value) && required.every(key => Object.hasOwn(value, key)) &&
+  Object.keys(value).every(key => required.includes(key) || optional.includes(key));
+const identity = value => JSON.stringify([value.taskPath, value.className, value.name]);
+function matchesPath(pattern, path) {
+  const expression = pattern.replace(/[.+^${}()|[\]\\]/g, "\\$&")
+    .replace(/\*\*/g, "\u0001").replace(/\*/g, "[^/]*").replace(/\?/g, "[^/]")
+    .replace(/\u0001\//g, "(?:.*/)?").replace(/\u0001/g, ".*");
+  return new RegExp(`^${expression}${pattern.endsWith("/") ? ".*" : ""}$`).test(path);
+}
+function validateContract(contract, config) {
+  assert(fields(contract, ["schemaVersion", "id", "title", "designApproved", "planPath", "ambiguityPolicy",
+    "maxFixLoops", "maxChangedFiles", "allowedPaths", "forbiddenPaths", "allowedWorkflowSkills",
+    "acceptanceCriteria", "nonGoals", "targetTests", "deviceTestsRequired", "testPolicy", "verification"],
+    contract.schemaVersion === 8 ? ["testPolicyReason", "recovery", "execution", "stageRecovery", "continuity"] : contract.schemaVersion === 7 ? ["testPolicyReason", "recovery", "execution", "stageRecovery"] : contract.schemaVersion === 6 ? ["testPolicyReason", "recovery", "execution"] : contract.schemaVersion === 5 ? ["testPolicyReason", "recovery"] : ["testPolicyReason"]), "Invalid inventory contract fields");
+  assert([4, 5, 6, 7, 8].includes(contract.schemaVersion) && /^TASK-[A-Z0-9-]+$/.test(contract.id), "Invalid inventory contract identity");
+  if (contract.schemaVersion >= 5) require("./recovery.cjs").validateRecoveryPolicy(contract.recovery);
+  if (contract.schemaVersion >= 6) require("./recovery.cjs").validateExecutionPolicy(contract.execution);
+  if (contract.schemaVersion >= 7) require("./recovery.cjs").validateStageRecoveryPolicy(contract.stageRecovery);
+  if (contract.schemaVersion === 8) {
+    const policy = contract.continuity;
+    assert(fields(policy, ["version", "isolatedAutoIntegration", "planningRefresh", "planningInputs"]) && policy.version === 1 &&
+      typeof policy.isolatedAutoIntegration === "boolean" && ["reject", "completedQueueTasks"].includes(policy.planningRefresh), "Invalid continuity policy");
+    assert(Array.isArray(policy.planningInputs) && policy.planningInputs.length > 0 && policy.planningInputs.length <= 128 &&
+      new Set(policy.planningInputs).size === policy.planningInputs.length && policy.planningInputs.every(p =>
+        typeof p === "string" && p.length <= 512 && /^[A-Za-z0-9._/?*-]+$/.test(p) &&
+        !p.startsWith("/") && !p.includes("..") && p.split("/").every(part => part && part !== "." && part !== ".git")), "Invalid planning inputs");
+  }
+  assert(text(contract.title) && contract.designApproved === true && contract.ambiguityPolicy === "BLOCKED", "An approved bounded design is required");
+  assert(contract.planPath === `docs/plans/${contract.id}.md`, "Invalid task plan path");
+  assert(Number.isInteger(contract.maxFixLoops) && contract.maxFixLoops >= 0 && contract.maxFixLoops <= 1, "Invalid fix-loop limit");
+  assert(Number.isInteger(contract.maxChangedFiles) && contract.maxChangedFiles >= 1 && contract.maxChangedFiles <= 12, "Invalid file limit");
+  for (const key of ["allowedPaths", "forbiddenPaths", "acceptanceCriteria", "nonGoals"])
+    assert(Array.isArray(contract[key]) && contract[key].length > 0 && contract[key].every(text), `Invalid ${key}`);
+  for (const path of [...contract.allowedPaths, ...contract.forbiddenPaths])
+    assert(/^[A-Za-z0-9._/?*-]+$/.test(path) && !path.startsWith("/") && !path.includes(".."), "Unsafe contract path");
+  for (const path of config.protectedPaths) {
+    assert(!contract.allowedPaths.some(pattern => matchesPath(pattern, path)), `Allowed paths overlap protected path: ${path}`);
+    assert(contract.forbiddenPaths.some(pattern => matchesPath(pattern, path)), `Forbidden paths must cover: ${path}`);
+  }
+  assert(JSON.stringify(contract.allowedWorkflowSkills) === JSON.stringify(["test-driven-development", "systematic-debugging", "verification-before-completion"].map(name => `android-orchestrator-${name}`)), "Invalid workflow skill allowlist");
+  assert(typeof contract.deviceTestsRequired === "boolean" && contract.testPolicy === "required", "Inventory verification requires tests");
+  assert(!contract.deviceTestsRequired || config.gradleVerification.deviceTestTasks?.length > 0,
+    "Device tests are required by this contract but no device test tasks are available");
+  assert(config.unitTestsEnabled !== false, "Inventory verification requires unitTestsEnabled=true");
+  assert(!Object.hasOwn(contract, "testPolicyReason") || typeof contract.testPolicyReason === "string", "Invalid test policy reason");
+  assert(Array.isArray(contract.targetTests) && contract.targetTests.length > 0, "Focused tests are required");
+  const targets = new Set();
+  for (const target of contract.targetTests) {
+    assert(fields(target, ["gradleTask", "filter"]) && /^(?::[A-Za-z0-9_.-]+)+$/.test(target.gradleTask), "Inventory verification requires a fully qualified Test task path");
+    assert(config.gradleVerification.focusedTestTasks.includes(target.gradleTask), "Focused test task is not configured");
+    assert(typeof target.filter === "string" && /^[A-Za-z0-9_.#$*-]+$/.test(target.filter), "Unsafe test filter");
+    const key = JSON.stringify([target.gradleTask, target.filter]);
+    assert(!targets.has(key), "Focused test targets must be unique"); targets.add(key);
+  }
+  const verification = contract.verification;
+  assert(fields(verification, ["version", "maxPreparationFixes", "cases", "inventory"], ["criteriaEvidence", "supplementalTests"]) && verification.version === 2, "Inventory verification requires verification version 2 with explicit inventory policy");
+  if (Object.hasOwn(verification, "supplementalTests")) assert(fields(verification.supplementalTests, ["mode", "maxRevisions"]) &&
+    verification.supplementalTests.mode === "baselinePassingNewFiles" && verification.supplementalTests.maxRevisions === 1,
+    "Supplemental tests require explicit baselinePassingNewFiles authorization and one revision");
+  assert(Number.isInteger(verification.maxPreparationFixes) && verification.maxPreparationFixes >= 0 && verification.maxPreparationFixes <= 1, "Invalid preparation retry budget");
+  const policy = verification.inventory;
+  assert(fields(policy, ["mode", "existingSkips", "emptyBaseline"]) && policy.mode === "focusedBaseline" &&
+    ["reject", "preserve"].includes(policy.existingSkips) && ["reject", "allow"].includes(policy.emptyBaseline), "Invalid inventory coverage, existing-skip or empty-baseline policy");
+  assert(Array.isArray(verification.cases) && verification.cases.length > 0, "Behavior cases are required");
+  const ids = new Set(), identities = new Set();
+  for (const item of verification.cases) {
+    assert(fields(item, ["id", "criterion", "intent", "before", "after", "source", "test"], ["expectedFailure"]), "Invalid behavior case fields");
+    assert(typeof item.id === "string" && /^[A-Z][A-Z0-9-]{2,63}$/.test(item.id) && !ids.has(item.id), "Behavior IDs must be unique and stable"); ids.add(item.id);
+    assert(Number.isInteger(item.criterion) && item.criterion >= 1 && item.criterion <= contract.acceptanceCriteria.length && item.after === "pass", "Invalid behavior criterion or GREEN expectation");
+    assert(fields(item.test, ["target", "className", "name"]) && Number.isInteger(item.test.target) && item.test.target >= 0 &&
+      item.test.target < contract.targetTests.length && text(item.test.className) && text(item.test.name), "Invalid exact test identity (include each parameter instance)");
+    const key = identity({ taskPath: contract.targetTests[item.test.target].gradleTask, ...item.test });
+    assert(!identities.has(key), "Behavior identities overlap across target filters"); identities.add(key);
+    if (item.intent === "preserve") assert(item.before === "pass" && ["existingTest", "baselineCapture", "measuredFact"].includes(item.source) && !Object.hasOwn(item, "expectedFailure"), "Preserved behavior must pass on a measured baseline");
+    else if (item.intent === "change") assert(item.before === "fail" && item.source === "userRequirement" && object(item.expectedFailure), "Changed behavior requires an approved expected failure");
+    else assert(item.intent === "observe" && item.before === "observe" && ["userRequirement", "existingTest", "baselineCapture", "measuredFact"].includes(item.source), "Invalid observed behavior");
+    if (Object.hasOwn(item, "expectedFailure")) {
+      const failure = item.expectedFailure;
+      const origin = failure.origin;
+      const structuredOrigin = fields(origin, ["className", "methodName"], ["fileName", "lineNumber"]) &&
+        text(origin.className) && /^[A-Za-z_$][A-Za-z0-9_.$]*$/.test(origin.className) && text(origin.methodName) &&
+        !/[().]/.test(origin.methodName) && (!Object.hasOwn(origin, "fileName") || text(origin.fileName) && !/[\\/]/.test(origin.fileName)) &&
+        (!Object.hasOwn(origin, "lineNumber") || Object.hasOwn(origin, "fileName") && Number.isSafeInteger(origin.lineNumber) && origin.lineNumber > 0);
+      assert(fields(failure, ["type", "origin"], ["messageIncludes"]) && text(failure.type) &&
+        (structuredOrigin || text(origin) && origin.length >= 12) &&
+        (!Object.hasOwn(failure, "messageIncludes") || (text(failure.messageIncludes) && failure.messageIncludes.length >= 3)), "Invalid expected failure");
+    }
+  }
+  assert(verification.cases.some(item => item.intent === "change"), "At least one changed behavior must provide genuine RED");
+  const covered = new Set(verification.cases.map(item => item.criterion));
+  if (Object.hasOwn(verification, "criteriaEvidence")) {
+    assert(Array.isArray(verification.criteriaEvidence), "Invalid criterion evidence mapping");
+    const explicit = new Set();
+    for (const evidence of verification.criteriaEvidence) {
+      assert(fields(evidence, ["criterion", "kind", "references"]) && Number.isInteger(evidence.criterion) &&
+        evidence.criterion >= 1 && evidence.criterion <= contract.acceptanceCriteria.length && !explicit.has(evidence.criterion), "Invalid or duplicate criterion evidence");
+      explicit.add(evidence.criterion);
+      assert(Array.isArray(evidence.references) && evidence.references.length > 0 && new Set(evidence.references).size === evidence.references.length, "Criterion evidence requires unique references");
+      const available = evidence.kind === "behavior" ? [...ids] : evidence.kind === "build" ? config.gradleVerification.assembleTasks :
+        evidence.kind === "lint" && config.lintEnabled === true ? config.gradleVerification.lintTasks :
+        evidence.kind === "device" && contract.deviceTestsRequired ? config.gradleVerification.deviceTestTasks : undefined;
+      assert(Array.isArray(available) && evidence.references.every(ref => available.includes(ref)), "Criterion evidence must reference approved behavior cases or mandatory enabled verification tasks");
+      covered.add(evidence.criterion);
+    }
+  }
+  assert(contract.acceptanceCriteria.every((_, index) => covered.has(index + 1)), "Every acceptance criterion requires an explicit behavior or verification evidence mapping");
+  assert(!/replace with|TASK-EXAMPLE|\btodo\b|\btbd\b|placeholder/i.test(JSON.stringify(contract)), "Contract contains placeholders");
+}
+module.exports = { validateContract, identity, matchesPath, assert };
+if (require.main === module) {
+  try { const fs = require("node:fs"); validateContract(JSON.parse(fs.readFileSync(process.argv[2], "utf8")), JSON.parse(fs.readFileSync(process.argv[3], "utf8"))); }
+  catch (error) { console.error(error.message); process.exitCode = 1; }
+}