feat: inject runtime model identity into prompts

CHANGELOG.md

@@ -7,6 +7,7 @@ Docs: https://docs.openclaw.ai
 ### Changes
 
 - Plugins/ACPX: accept an optional `args` array in `agents.<name>` config so paths and flag values containing spaces stay intact when spawning ACP agent processes. Thanks @TheArchitectit and @BunsDev.
+- Agents: inject the current provider/model identity into system prompts, including configured prompt overrides and CLI hook prompt replacements, so agents can answer model-identity questions from the actual runtime selection.
 - Plugins/CLI: add the optional bundled `oc-path` plugin, providing `openclaw path` for surgical `oc://` access to markdown, JSONC, and JSONL workspace files.
 - Plugins/SDK: add unified model catalog registration for text, image, video, and music providers, including `providerCatalogEntry` manifests, shared media list help, live catalog caching, and per-model video capability overlays.
 - CLI: make parser, startup, config, guardrail, channel, agent, task, session, and MCP failures explain what happened and point to the next recovery command.

src/agents/cli-runner/prepare.test.ts

@@ -273,7 +273,9 @@ describe("shouldSkipLocalCliCredentialEpoch", () => {
       });
 
       expect(context.params.prompt).toBe("history:2\n\nlatest ask");
-      expect(context.systemPrompt).toBe("prepend system\n\nhook system\n\nappend system");
+      expect(context.systemPrompt).toBe(
+        "prepend system\n\nhook system\n\nappend system\n\nCurrent model identity: test-cli/test-model. If asked what model you are, answer with this value for the current run.",
+      );
       expect(hookRunner.runBeforePromptBuild).toHaveBeenCalledWith(
         {
           prompt: "latest ask",
@@ -441,7 +443,7 @@ describe("shouldSkipLocalCliCredentialEpoch", () => {
 
       expect(context.params.prompt).toBe("prompt prepend\n\nlegacy prepend\n\nlatest ask");
       expect(context.systemPrompt).toBe(
-        "prompt prepend system\n\nlegacy prepend system\n\nprompt system\n\nprompt append system\n\nlegacy append system",
+        "prompt prepend system\n\nlegacy prepend system\n\nprompt system\n\nprompt append system\n\nlegacy append system\n\nCurrent model identity: test-cli/test-model. If asked what model you are, answer with this value for the current run.",
       );
       expect(hookRunner.runBeforePromptBuild).toHaveBeenCalledOnce();
       expect(hookRunner.runBeforeAgentStart).toHaveBeenCalledOnce();
@@ -475,7 +477,9 @@ describe("shouldSkipLocalCliCredentialEpoch", () => {
       });
 
       expect(context.params.prompt).toBe("latest ask");
-      expect(context.systemPrompt).toBe("base extra system");
+      expect(context.systemPrompt).toBe(
+        "base extra system\n\nCurrent model identity: test-cli/test-model. If asked what model you are, answer with this value for the current run.",
+      );
       expect(context.systemPrompt).not.toContain("hook exploded");
       expect(hookRunner.runBeforePromptBuild).toHaveBeenCalledOnce();
     } finally {
@@ -570,7 +574,9 @@ describe("shouldSkipLocalCliCredentialEpoch", () => {
         config: createCliBackendConfig(),
       });
 
-      expect(context.systemPrompt).toBe("active video task\n\nhook prepend system\n\nhook system");
+      expect(context.systemPrompt).toBe(
+        "active video task\n\nhook prepend system\n\nhook system\n\nCurrent model identity: test-cli/test-model. If asked what model you are, answer with this value for the current run.",
+      );
       expect(mockBuildActiveVideoGenerationTaskPromptContextForSession).toHaveBeenCalledWith(
         "agent:main:test",
       );

src/agents/cli-runner/prepare.ts

@@ -43,6 +43,7 @@ import { applyPluginTextReplacements } from "../plugin-text-transforms.js";
 import { resolveSkillsPromptForRun } from "../skills.js";
 import { resolveSystemPromptOverride } from "../system-prompt-override.js";
 import { buildSystemPromptReport } from "../system-prompt-report.js";
+import { appendModelIdentitySystemPrompt } from "../system-prompt.js";
 import { redactRunIdentifier, resolveRunWorkspaceDir } from "../workspace-run.js";
 import { prepareCliBundleMcpConfig } from "./bundle-mcp.js";
 import { buildSystemPrompt, normalizeCliModel } from "./helpers.js";
@@ -417,7 +418,10 @@ export async function prepareCliRunContext(
         }),
         prompt: preparedPrompt,
       });
-  systemPrompt = applyPluginTextReplacements(systemPrompt, backendResolved.textTransforms?.input);
+  systemPrompt = appendModelIdentitySystemPrompt({
+    systemPrompt: applyPluginTextReplacements(systemPrompt, backendResolved.textTransforms?.input),
+    model: modelDisplay,
+  });
   const systemPromptReport = buildSystemPromptReport({
     source: "run",
     generatedAt: Date.now(),

src/agents/pi-embedded-runner/run/attempt-system-prompt.test.ts

@@ -58,6 +58,7 @@ describe("buildAttemptSystemPrompt", () => {
     });
 
     expect(result.systemPrompt).toContain("Custom override prompt.");
+    expect(result.systemPrompt).toContain("Current model identity: openai/gpt-5.5.");
     expect(result.systemPrompt).toContain("## Bootstrap Pending");
     expect(result.systemPrompt).toContain("BOOTSTRAP.md is included below in Project Context");
     expect(result.systemPrompt).toContain("## Bootstrap Context Notice");
@@ -96,6 +97,7 @@ describe("buildAttemptSystemPrompt", () => {
     });
 
     expect(result.systemPrompt).toContain("Custom override prompt.");
+    expect(result.systemPrompt).toContain("Current model identity: openai/gpt-5.5.");
     expect(result.systemPrompt).toContain("## Subagent Context");
     expect(result.systemPrompt).toContain("RUN_MODE_TASK_77950");
   });

src/agents/pi-embedded-runner/run/attempt-system-prompt.ts

@@ -1,6 +1,9 @@
 import type { OpenClawConfig } from "../../../config/types.openclaw.js";
 import type { ProviderTransformSystemPromptContext } from "../../../plugins/types.js";
-import { appendAgentBootstrapSystemPromptSupplement } from "../../system-prompt.js";
+import {
+  appendAgentBootstrapSystemPromptSupplement,
+  appendModelIdentitySystemPrompt,
+} from "../../system-prompt.js";
 import { buildEmbeddedSystemPrompt, createSystemPromptOverride } from "../system-prompt.js";
 
 type EmbeddedSystemPromptParams = Parameters<typeof buildEmbeddedSystemPrompt>[0];
@@ -48,15 +51,18 @@ export function buildAttemptSystemPrompt(
   params: BuildAttemptSystemPromptParams,
 ): AttemptSystemPrompt {
   const baseSystemPrompt = params.systemPromptOverrideText
-    ? appendRuntimeExtraSystemPrompt({
-        systemPrompt: appendAgentBootstrapSystemPromptSupplement({
-          systemPrompt: params.systemPromptOverrideText,
-          bootstrapMode: params.embeddedSystemPrompt.bootstrapMode,
-          bootstrapTruncationNotice: params.embeddedSystemPrompt.bootstrapTruncationNotice,
-          contextFiles: params.embeddedSystemPrompt.contextFiles,
+    ? appendModelIdentitySystemPrompt({
+        systemPrompt: appendRuntimeExtraSystemPrompt({
+          systemPrompt: appendAgentBootstrapSystemPromptSupplement({
+            systemPrompt: params.systemPromptOverrideText,
+            bootstrapMode: params.embeddedSystemPrompt.bootstrapMode,
+            bootstrapTruncationNotice: params.embeddedSystemPrompt.bootstrapTruncationNotice,
+            contextFiles: params.embeddedSystemPrompt.contextFiles,
+          }),
+          extraSystemPrompt: params.embeddedSystemPrompt.extraSystemPrompt,
+          promptMode: params.embeddedSystemPrompt.promptMode,
         }),
-        extraSystemPrompt: params.embeddedSystemPrompt.extraSystemPrompt,
-        promptMode: params.embeddedSystemPrompt.promptMode,
+        model: params.embeddedSystemPrompt.runtimeInfo.model,
       })
     : buildEmbeddedSystemPrompt(params.embeddedSystemPrompt);
 

src/agents/pi-embedded-runner/run/attempt.ts

@@ -165,6 +165,7 @@ import {
 import { resolveSystemPromptOverride } from "../../system-prompt-override.js";
 import { buildSystemPromptParams } from "../../system-prompt-params.js";
 import { buildSystemPromptReport } from "../../system-prompt-report.js";
+import { appendModelIdentitySystemPrompt } from "../../system-prompt.js";
 import { resolveAgentTimeoutMs } from "../../timeout.js";
 import {
   buildEmptyExplicitToolAllowlistError,
@@ -2717,6 +2718,14 @@ export async function runEmbeddedAttempt(
             );
           }
         }
+        const modelAwareSystemPrompt = appendModelIdentitySystemPrompt({
+          systemPrompt: systemPromptText,
+          model: runtimeInfo.model,
+        });
+        if (modelAwareSystemPrompt !== systemPromptText) {
+          applySystemPromptOverrideToSession(activeSession, modelAwareSystemPrompt);
+          systemPromptText = modelAwareSystemPrompt;
+        }
 
         if (cacheObservabilityEnabled) {
           const cacheObservation = beginPromptCacheObservation({

src/agents/system-prompt.test.ts

@@ -101,6 +101,20 @@ describe("buildAgentSystemPrompt", () => {
     expect(tokenA).not.toBe(tokenB);
   });
 
+  it("injects the current model identity into the runtime prompt", () => {
+    const prompt = buildAgentSystemPrompt({
+      workspaceDir: "/tmp/openclaw",
+      runtimeInfo: {
+        agentId: "main",
+        model: "openai/gpt-5.5",
+      },
+    });
+
+    expect(prompt).toContain(
+      "Current model identity: openai/gpt-5.5. If asked what model you are, answer with this value for the current run.",
+    );
+  });
+
   it("omits extended sections in minimal prompt mode", () => {
     const prompt = buildAgentSystemPrompt({
       workspaceDir: "/tmp/openclaw",

src/agents/system-prompt.ts

@@ -550,6 +550,51 @@ function formatFullAccessBlockedReason(reason?: EmbeddedFullAccessBlockedReason)
   }
   return "runtime constraints";
 }
+
+const MODEL_IDENTITY_PREFIX = "Current model identity:";
+
+export function buildModelIdentityPromptLine(model?: string): string | undefined {
+  const trimmed = model?.trim();
+  if (!trimmed) {
+    return undefined;
+  }
+  return `${MODEL_IDENTITY_PREFIX} ${trimmed}. If asked what model you are, answer with this value for the current run.`;
+}
+
+export function appendModelIdentitySystemPrompt(params: {
+  systemPrompt: string;
+  model?: string;
+}): string {
+  const line = buildModelIdentityPromptLine(params.model);
+  if (!line) {
+    return params.systemPrompt;
+  }
+
+  let replaced = false;
+  const nextLines = params.systemPrompt
+    .split(/\r?\n/u)
+    .filter((candidate) => {
+      if (!candidate.trimStart().startsWith(MODEL_IDENTITY_PREFIX)) {
+        return true;
+      }
+      if (replaced) {
+        return false;
+      }
+      replaced = true;
+      return true;
+    })
+    .map((candidate) =>
+      candidate.trimStart().startsWith(MODEL_IDENTITY_PREFIX) ? line : candidate,
+    );
+
+  if (replaced) {
+    return nextLines.join("\n");
+  }
+
+  const base = params.systemPrompt.trimEnd();
+  return base ? `${base}\n\n${line}` : line;
+}
+
 export function buildAgentSystemPrompt(params: {
   workspaceDir: string;
   defaultThinkLevel?: ThinkLevel;
@@ -757,6 +802,7 @@ export function buildAgentSystemPrompt(params: {
   const skillsPrompt = params.skillsPrompt?.trim();
   const heartbeatPrompt = params.heartbeatPrompt?.trim();
   const runtimeInfo = params.runtimeInfo;
+  const modelIdentityLine = buildModelIdentityPromptLine(runtimeInfo?.model);
   const runtimeChannel = normalizeOptionalLowercaseString(runtimeInfo?.channel);
   const runtimeCapabilities = runtimeInfo?.capabilities ?? [];
   const runtimeCapabilitiesLower = new Set(
@@ -816,7 +862,9 @@ export function buildAgentSystemPrompt(params: {
 
   // For "none" mode, return just the basic identity line
   if (promptMode === "none") {
-    return "You are a personal assistant running inside OpenClaw.";
+    return ["You are a personal assistant running inside OpenClaw.", modelIdentityLine]
+      .filter(Boolean)
+      .join("\n");
   }
 
   const contextFiles = params.contextFiles ?? [];
@@ -1174,6 +1222,7 @@ export function buildAgentSystemPrompt(params: {
   lines.push(
     "## Runtime",
     buildRuntimeLine(runtimeInfo, runtimeChannel, runtimeCapabilities, params.defaultThinkLevel),
+    ...(modelIdentityLine ? [modelIdentityLine] : []),
     `Reasoning: ${reasoningLevel} (hidden unless on/stream). Toggle /reasoning; /status shows Reasoning when enabled.`,
   );