@principles/host-runtime 0.3.11 → 0.3.12

package/dist/internalization-consumer-cycle.js

@@ -21,7 +21,7 @@
  * env getter. Everything else is the existing production path.
  */
 import { createHash } from 'node:crypto';
-import { createRuntimeStateHandle, InternalizationOrchestrator, DreamerRunner, PhilosopherRunner, ScribeRunner, ArtificerRunner, EvaluatorRunner, RolloutReviewerRunner, DefaultDreamerValidator, DefaultPhilosopherValidator, DefaultScribeValidator, DefaultArtificerValidator, DefaultEvaluatorValidator, DefaultRolloutReviewerValidator, PiAiRuntimeAdapter, L2AgentLoopAdapter, buildL2PrincipleReaderFromLedger, OpenClawCliRuntimeAdapter, storeEmitter, createProductionGateDeps, resolveRuntimeConfigFromPdConfig, isRuntimeConfigError, computeConsumerDecision, FULL_CHAIN_CONSUMER_RUNNER_KINDS, DEFAULT_CONSUMER_RUNNER_KINDS, InternalizationQueueReadModel, MVP_CORE_TASK_KINDS, SqliteConnection, SqliteReconciliationCursorStore, SUCCEEDED_TRANSITIONS_SCOPE, 
+import { createRuntimeStateHandle, InternalizationOrchestrator, DreamerRunner, PhilosopherRunner, ScribeRunner, ArtificerRunner, EvaluatorRunner, RolloutReviewerRunner, DefaultDreamerValidator, DefaultPhilosopherValidator, DefaultScribeValidator, DefaultArtificerValidator, DefaultEvaluatorValidator, DefaultRolloutReviewerValidator, PiAiRuntimeAdapter, L2AgentLoopAdapter, buildL2PrincipleReaderFromLedger, OpenClawCliRuntimeAdapter, storeEmitter, createProductionGateDeps, resolveRuntimeConfigFromPdConfig, resolveRuntimeConfigForAgent, AGENT_NAME_FOR_TASK_KIND, isRuntimeConfigError, computeConsumerDecision, FULL_CHAIN_CONSUMER_RUNNER_KINDS, DEFAULT_CONSUMER_RUNNER_KINDS, InternalizationQueueReadModel, MVP_CORE_TASK_KINDS, SqliteConnection, SqliteReconciliationCursorStore, SUCCEEDED_TRANSITIONS_SCOPE, 
 // PRI-714: resolve outputLanguage from the same effective config the
 // runners already receive (EP-07: canonical resolved value, not raw input).
 resolveOutputLanguage, } from '@principles/core/runtime-v2';
@@ -240,13 +240,50 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
         // Advance the configured runner kinds in priority order (dreamer first,
         // then philosopher→…→rollout_reviewer under full-chain scope). Lease the
         // first ready task whose dependencies are satisfied.
+        //
+        // PRI-719 (EP002-R2 F4): the runtime profile is resolved PER KIND, before
+        // wakeOnce — `internalAgents.agents[kind].runtimeProfile` (fallback
+        // defaultRuntime) now actually governs the stage that runs on it. Before,
+        // the whole chain shared the diagnostician binding and every per-agent
+        // profile declaration was silently ignored. A kind whose profile cannot
+        // resolve is skipped observably (no lease taken for it).
         let wakeResult = null;
         let lastSkipDecision = 'no_ready_tasks';
         let lastSkipReason;
+        let taskRuntimeConfig = null;
         for (const kind of decision.runnerKinds) {
+            const agentName = AGENT_NAME_FOR_TASK_KIND[kind];
+            if (agentName === undefined) {
+                lastSkipDecision = 'runtime_config_error';
+                lastSkipReason = `no internal agent mapping for task kind '${kind}'`;
+                continue;
+            }
+            const kindRuntime = resolveRuntimeConfigForAgent(configResult.effective, agentName, {
+                getEnvVar: (name) => envGetter(name),
+                // Consumer stage scope is the internalization_full_chain flag, not
+                // agents[kind].enabled (the shipped default disables
+                // philosopher/evaluator/rolloutReviewer yet full-chain runs them) —
+                // see ResolveRuntimeConfigForAgentOptions.
+                ignoreAgentEnabled: true,
+            });
+            if (isRuntimeConfigError(kindRuntime)) {
+                // rc-9: the degradation is observable per kind; other kinds still run.
+                emitEvent('INTERNALIZATION_CONSUMER_SKIP', JSON.stringify({
+                    reason: 'runtime_config_error',
+                    taskKind: kind,
+                    agentName,
+                    message: kindRuntime.message,
+                    nextAction: kindRuntime.nextAction,
+                }));
+                logger.warn(`[PD:${logLabel}] Runtime config error for ${kind} (agent ${agentName}): ${kindRuntime.message}`);
+                lastSkipDecision = 'runtime_config_error';
+                lastSkipReason = kindRuntime.message;
+                continue;
+            }
             const candidate = await orchestrator.wakeOnce(kind);
             if (candidate.decision === 'would_lease') {
                 wakeResult = candidate;
+                taskRuntimeConfig = kindRuntime;
                 break;
             }
             // Keep decision/reason paired across iterations so the final SKIP log
@@ -254,7 +291,7 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
             lastSkipDecision = candidate.decision;
             lastSkipReason = candidate.decision === 'no_ready_tasks' ? candidate.reason : undefined;
         }
-        if (!wakeResult) {
+        if (!wakeResult || taskRuntimeConfig === null) {
             // A: 预算由周期 finally 统一执行 (每周期恰一次,含 backlog 场景)
             const skipPayload = { decision: lastSkipDecision };
             if (lastSkipReason) {
@@ -264,8 +301,11 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
             logger.info(`[PD:${logLabel}] No task to consume: ${lastSkipDecision}`);
             return { ran: false, skipReason: lastSkipDecision };
         }
+        // PRI-719: everything downstream reads the LEASED TASK's resolved config —
+        // the kind-specific profile, not the shared gate resolution.
+        const taskRuntimeKind = taskRuntimeConfig.runtimeKind;
         let adapter;
-        if (runtimeKind === 'pi-ai') {
+        if (taskRuntimeKind === 'pi-ai') {
             // PRI-419: when l2_dreamer flag is on AND this is a dreamer task, route
             // through the L2 multi-turn agent loop. Non-dreamer runners always use PiAi.
             const l2Flag = loadFeatureFlagFromConfig(workspaceDir, 'l2_dreamer');
@@ -275,14 +315,14 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
                     logger: { warn: (msg) => logger.warn(msg) },
                 });
                 adapter = new L2AgentLoopAdapter({
-                    provider: runtimeConfigResult.provider ?? 'openai',
-                    model: runtimeConfigResult.model ?? 'gpt-4o',
-                    apiKeyEnv: runtimeConfigResult.apiKeyEnv ?? 'OPENAI_API_KEY',
-                    baseUrl: runtimeConfigResult.baseUrl,
+                    provider: taskRuntimeConfig.provider ?? 'openai',
+                    model: taskRuntimeConfig.model ?? 'gpt-4o',
+                    apiKeyEnv: taskRuntimeConfig.apiKeyEnv ?? 'OPENAI_API_KEY',
+                    baseUrl: taskRuntimeConfig.baseUrl,
                     workspace: workspaceDir,
-                    totalBudgetMs: runtimeConfigResult.timeoutMs,
+                    totalBudgetMs: taskRuntimeConfig.timeoutMs,
                     // PRI-633: profile systemPrompt rides as the append layer.
-                    ...(runtimeConfigResult.systemPrompt ? { systemPrompt: runtimeConfigResult.systemPrompt } : {}),
+                    ...(taskRuntimeConfig.systemPrompt ? { systemPrompt: taskRuntimeConfig.systemPrompt } : {}),
                 }, {
                     artifactReader: {
                         // Explicit adapter: PIArtifactRecord → PdL2ArtifactReader. The store returns
@@ -301,27 +341,27 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
             }
             else {
                 adapter = new PiAiRuntimeAdapter({
-                    provider: runtimeConfigResult.provider ?? 'openai',
-                    model: runtimeConfigResult.model ?? 'gpt-4o',
-                    apiKeyEnv: runtimeConfigResult.apiKeyEnv ?? 'OPENAI_API_KEY',
-                    maxRetries: runtimeConfigResult.maxRetries,
-                    maxTokens: runtimeConfigResult.maxTokens,
-                    timeoutMs: runtimeConfigResult.timeoutMs,
-                    baseUrl: runtimeConfigResult.baseUrl,
+                    provider: taskRuntimeConfig.provider ?? 'openai',
+                    model: taskRuntimeConfig.model ?? 'gpt-4o',
+                    apiKeyEnv: taskRuntimeConfig.apiKeyEnv ?? 'OPENAI_API_KEY',
+                    maxRetries: taskRuntimeConfig.maxRetries,
+                    maxTokens: taskRuntimeConfig.maxTokens,
+                    timeoutMs: taskRuntimeConfig.timeoutMs,
+                    baseUrl: taskRuntimeConfig.baseUrl,
                     workspace: workspaceDir,
                     // PRI-633: profile systemPrompt rides as the append layer.
-                    ...(runtimeConfigResult.systemPrompt ? { systemPrompt: runtimeConfigResult.systemPrompt } : {}),
+                    ...(taskRuntimeConfig.systemPrompt ? { systemPrompt: taskRuntimeConfig.systemPrompt } : {}),
                 });
             }
         }
-        else if (runtimeKind === 'openclaw-cli') {
+        else if (taskRuntimeKind === 'openclaw-cli') {
             adapter = new OpenClawCliRuntimeAdapter({
-                runtimeMode: runtimeConfigResult.openclawMode ?? 'default',
+                runtimeMode: taskRuntimeConfig.openclawMode ?? 'default',
                 workspaceDir: workspaceDir,
             });
         }
         else {
-            throw new Error(`Unsupported runtime kind resolved for auto-consumer: ${runtimeKind}`);
+            throw new Error(`Unsupported runtime kind resolved for auto-consumer: ${taskRuntimeKind}`);
         }
         const { taskId } = wakeResult;
         const { taskKind } = wakeResult;
@@ -340,9 +380,9 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
         // same effective config (EP-07 canonical value).
         const runnerOptions = {
             owner,
-            runtimeKind,
+            runtimeKind: taskRuntimeKind,
             effectiveConfig: configResult.effective,
-            timeoutMs: runtimeConfigResult.timeoutMs,
+            timeoutMs: taskRuntimeConfig.timeoutMs,
             outputLanguage: resolveOutputLanguage(configResult.effective.config.principles?.outputLanguage).outputLanguage,
         };
         // PRI-634 A3: workspace-scoped telemetry sink for the evaluator runner.
@@ -423,10 +463,18 @@ export async function runInternalizationConsumerCycle(workspaceDir, ports) {
                 logger.warn(`[PD:${logLabel}] No consumer runner for task kind '${taskKind}'; skipping. Advance manually: pd runtime internalization run-once --runner ${taskKind}`);
                 return { ran: false, skipReason: 'no_runner_for_kind', taskKind };
         }
-        logger.info(`[PD:${logLabel}] Running ${taskKind} task: ${taskId}`);
+        // PRI-719: run evidence answers "which declared profile/provider/model
+        // actually executed this task" (declared == executed contract).
+        const runtimeEvidence = {
+            ...(taskRuntimeConfig.runtimeProfileId !== undefined ? { runtimeProfileId: taskRuntimeConfig.runtimeProfileId } : {}),
+            ...(taskRuntimeConfig.provider !== undefined ? { provider: taskRuntimeConfig.provider } : {}),
+            ...(taskRuntimeConfig.model !== undefined ? { model: taskRuntimeConfig.model } : {}),
+        };
+        logger.info(`[PD:${logLabel}] Running ${taskKind} task: ${taskId}${Object.keys(runtimeEvidence).length > 0 ? ` (runtime: ${JSON.stringify(runtimeEvidence)})` : ''}`);
         emitEvent('INTERNALIZATION_CONSUMER_RUN', JSON.stringify({
             taskId,
             taskKind,
+            ...runtimeEvidence,
         }));
         let runResult;
         try {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@principles/host-runtime",
-  "version": "0.3.11",
+  "version": "0.3.12",
   "description": "Shared host-neutral orchestration for Principles Disciple MVP-Core hook paths.",
   "type": "module",
   "main": "./dist/index.js",