diff --git a/docs/plans/state.json b/docs/plans/state.json
index 2955b5f..4ab7bb8 100644
--- a/docs/plans/state.json
+++ b/docs/plans/state.json
@@ -1424,6 +1424,15 @@
               "src/cli/skills.test.ts"
             ],
             "test_status": "pnpm typecheck + pnpm test:run src/cli/skills.test.ts + pnpm test:run + pnpm lint (warnings only, 0 errors) + pnpm build passing"
+          },
+          "installer_runner_interface_noop": {
+            "status": "completed",
+            "description": "Added a pluggable installer command runner interface with policy-gated dispatch and a default no-op runner, preserving execution-disabled behavior while preparing a future real runner",
+            "files_modified": [
+              "src/cli/skills.ts",
+              "src/cli/skills.test.ts"
+            ],
+            "test_status": "pnpm typecheck + pnpm test:run src/cli/skills.test.ts + pnpm test:run + pnpm lint (warnings only, 0 errors) + pnpm build passing"
           }
         }
       }
@@ -1452,7 +1461,7 @@
   },
 
   "overall_progress": {
-    "total_test_count": 1537,
+    "total_test_count": 1540,
     "all_tests_passing": true,
     "p0_completion": "3/3 (100%)",
     "p1_completion": "4/4 (100%)",
@@ -1472,7 +1481,7 @@
     "gmail_auth_cli": "flynn gmail-auth command implemented with OAuth2 flow, doctor check, config routed to Telegram",
     "native_audio_support": "completed — smart routing for native audio (Gemini/OpenAI/GitHub) vs Whisper transcription fallback",
     "remaining_phases_completion": "Phase 1: 3/3 (100%) — context levels, command registry, memory structure. Phase 2: 2/2 (100%) — component registry, confidence routing. Phase 3: 2/2 (100%) — adaptive memory/compaction, truthfulness/autonomy hardening",
-    "next_up": "Skills infrastructure Phase 3: introduce a pluggable installer command runner interface behind the existing execution policy gates (default no-op)"
+    "next_up": "Skills infrastructure Phase 3: add structured per-step execution result envelopes for the future real runner while keeping no-op default"
   },
   "soul_md_and_cron_create": {
     "date": "2026-02-11",
diff --git a/src/cli/skills.test.ts b/src/cli/skills.test.ts
index 9577097..6622eb1 100644
--- a/src/cli/skills.test.ts
+++ b/src/cli/skills.test.ts
@@ -19,6 +19,8 @@ import {
   renderSkillInstallerExecutionStub,
   toSkillInstallerExecutionStubFromPreflight,
   evaluateInstallerExecutionPolicy,
+  runInstallerCommandsWithPolicy,
+  noOpSkillInstallerCommandRunner,
   runSkillInstallAction,
 } from './skills.js';
 import type { Skill } from '../skills/index.js';
@@ -296,6 +298,36 @@ describe('skills CLI helpers', () => {
     expect(policy.reason).toBe('execution_disabled');
   });
 
+  it('does not invoke command runner when policy disables execution', () => {
+    const runner = {
+      run: vi.fn((_commands: string[]) => ['should-not-run']),
+    };
+
+    const executed = runInstallerCommandsWithPolicy(
+      ['brew install jq'],
+      { confirmed: false, execution_enabled: false, reason: 'confirmation_required' },
+      runner,
+    );
+
+    expect(executed).toEqual([]);
+    expect(runner.run).not.toHaveBeenCalled();
+  });
+
+  it('supports pluggable command runner when policy enables execution', () => {
+    const runner = {
+      run: vi.fn((commands: string[]) => commands),
+    };
+
+    const executed = runInstallerCommandsWithPolicy(
+      ['brew install jq'],
+      { confirmed: true, execution_enabled: true, reason: 'execution_disabled' },
+      runner,
+    );
+
+    expect(executed).toEqual(['brew install jq']);
+    expect(runner.run).toHaveBeenCalledWith(['brew install jq']);
+  });
+
   it('summarizes refresh counts across status and tiers', () => {
     const summary = summarizeSkillsRefresh([
       buildSkill({ manifest: { name: 'a', description: 'a', version: '1.0.0', tier: 'bundled' } }),
@@ -436,6 +468,34 @@ describe('skills CLI helpers', () => {
     rmSync(root, { recursive: true, force: true });
   });
 
+  it('uses no-op command runner by default in install flow', () => {
+    const root = mkdtempSync(join(tmpdir(), 'flynn-skills-cli-'));
+    const sourceDir = join(root, 'source-skill');
+    const managedDir = join(root, 'managed');
+    mkdirSync(sourceDir, { recursive: true });
+    writeFileSync(join(sourceDir, 'SKILL.md'), '# Install Skill\nInstructions');
+    writeFileSync(
+      join(sourceDir, 'manifest.json'),
+      JSON.stringify({ name: 'install-skill', description: 'Install', version: '1.0.0' }),
+      'utf-8',
+    );
+
+    const installer = new SkillInstaller(managedDir);
+    const logSpy = vi.spyOn(console, 'log').mockImplementation(() => undefined);
+    const runnerSpy = vi.spyOn(noOpSkillInstallerCommandRunner, 'run');
+
+    const result = runSkillInstallAction(installer, sourceDir, { mode: 'install', asJson: true, confirmed: true });
+
+    expect(result.ok).toBe(true);
+    expect(runnerSpy).not.toHaveBeenCalled();
+    const payload = JSON.parse(String(logSpy.mock.calls[logSpy.mock.calls.length - 1]?.[0]));
+    expect(payload.execution.executed).toEqual([]);
+
+    runnerSpy.mockRestore();
+    logSpy.mockRestore();
+    rmSync(root, { recursive: true, force: true });
+  });
+
   it('requires --yes confirmation for uninstall helper', () => {
     const root = mkdtempSync(join(tmpdir(), 'flynn-skills-cli-'));
     const installer = new SkillInstaller(join(root, 'managed'));
diff --git a/src/cli/skills.ts b/src/cli/skills.ts
index 4f35356..ab48dce 100644
--- a/src/cli/skills.ts
+++ b/src/cli/skills.ts
@@ -59,6 +59,16 @@ export interface SkillInstallerExecutionPolicy {
   reason: SkillInstallerExecutionReason;
 }
 
+export interface SkillInstallerCommandRunner {
+  run(commands: string[]): string[];
+}
+
+export const noOpSkillInstallerCommandRunner: SkillInstallerCommandRunner = {
+  run(_commands: string[]): string[] {
+    return [];
+  },
+};
+
 export function toSkillListRows(skills: Skill[]): SkillListRow[] {
   return skills
     .map((skill) => ({
@@ -271,6 +281,18 @@ export function evaluateInstallerExecutionPolicy(opts: {
   };
 }
 
+export function runInstallerCommandsWithPolicy(
+  commands: string[],
+  policy: SkillInstallerExecutionPolicy,
+  runner: SkillInstallerCommandRunner,
+): string[] {
+  if (!policy.execution_enabled) {
+    return [];
+  }
+
+  return runner.run(commands);
+}
+
 export function renderSkillInstallerExecutionStub(view: SkillInstallerExecutionStubView): string {
   const lines: string[] = [
     `Installer execution stub for '${view.skill.name}' (${view.skill.tier}, v${view.skill.version})`,
@@ -359,7 +381,12 @@ export function installSkillFromDirectory(installer: SkillInstaller, sourcePath:
 export function runSkillInstallAction(
   installer: SkillInstaller,
   sourcePath: string,
-  opts: { mode: SkillInstallActionMode; asJson: boolean; confirmed: boolean },
+  opts: {
+    mode: SkillInstallActionMode;
+    asJson: boolean;
+    confirmed: boolean;
+    commandRunner?: SkillInstallerCommandRunner;
+  },
 ): { ok: true } | { ok: false; error: string } {
   const preflight = toSkillInstallPreflightView(sourcePath);
 
@@ -422,12 +449,18 @@ export function runSkillInstallAction(
         mode: 'install' as const,
         confirmed: installPolicy.confirmed,
         execution_enabled: installPolicy.execution_enabled,
-        executed: [],
+        executed: runInstallerCommandsWithPolicy([], installPolicy, opts.commandRunner ?? noOpSkillInstallerCommandRunner),
         reason: installPolicy.reason,
         wouldRun: [],
         skipped: [],
       };
 
+  execution.executed = runInstallerCommandsWithPolicy(
+    execution.wouldRun,
+    installPolicy,
+    opts.commandRunner ?? noOpSkillInstallerCommandRunner,
+  );
+
   if (opts.asJson) {
     console.log(
       JSON.stringify(