build(plugins): pin retained runtime and cover handoff error paths

integrations/openclaw/skill-loader.test.ts

@@ -131,6 +131,10 @@ describe("loadCompactTriagePrompt", () => {
     });
 
     expect(prompt).toContain("No automatic recall happens in manual mode.");
+    const fullPrompt = loadTriagePrompt({recall: {strategy: "manual"}});
+    expect(fullPrompt).toContain("No automatic recall happens in manual mode.");
+    expect(fullPrompt).toContain(SEARCH_GUIDANCE);
+    expect(fullPrompt).not.toContain("ALWAYS rewrite");
   });
 
   it("includes short config summaries and truncates oversized custom rules", () => {

integrations/openclaw/tests/handoff.test.ts

@@ -6,14 +6,14 @@ import {runNativeSession, runHandoffAction} from "../../agent-plugin-core/typesc
 import {registerHandoffCommand} from "../tools/handoff.ts";
 vi.mock("../../agent-plugin-core/typescript/src/handoff.ts", async original => ({...await original<typeof import("../../agent-plugin-core/typescript/src/handoff.ts")>(), runNativeSession: vi.fn(), runHandoffAction: vi.fn()}));
 let dir: string;
-beforeEach(async () => {vi.clearAllMocks(); dir = await mkdtemp(join(tmpdir(), "openclaw-handoff-"));});
+beforeEach(async () => {vi.resetAllMocks(); dir = await mkdtemp(join(tmpdir(), "openclaw-handoff-"));});
 afterEach(async () => {await rm(dir, {recursive: true, force: true});});
-function setup() {
+function setup(context: {workspaceDir?: string} = {workspaceDir: "/tmp/native-project"}) {
   const registerCommand = vi.fn();
   const registerTool = vi.fn();
   const hooks = new Map<string, any>();
   registerHandoffCommand({registerCommand, registerTool, on: (name: string, handler: any) => hooks.set(name, handler)} as any);
-  return {cmd: registerCommand.mock.calls[0][0], tool: registerTool.mock.calls[0][0]({workspaceDir: "/tmp/native-project"}), hooks};
+  return {cmd: registerCommand.mock.calls[0][0], tool: registerTool.mock.calls[0][0](context), hooks};
 }
 it("saves the trusted current OpenClaw transcript from a user-only command", async () => {
   const {cmd} = setup();
@@ -61,3 +61,58 @@ it("fails explicitly on unavailable context or resource errors", async () => {
   expect((await cmd.handler({sessionFile: "/tmp/native.jsonl"})).text).toContain("invalid native transcript");
   expect((await cmd.handler({args: "resume /tmp/shared.json", sessionFile: "/tmp/native.jsonl"})).text).toContain("identity is unavailable");
 });
+
+it("lists resources from the transcript project without queuing prompt context", async () => {
+  const {cmd, hooks} = setup();
+  const sessionFile = join(dir, "native.jsonl");
+  await writeFile(sessionFile, JSON.stringify({type: "session", cwd: dir}) + "\n");
+  vi.mocked(runHandoffAction).mockResolvedValue("Available shared resource");
+  expect(await cmd.handler({args: "list", sessionFile, sessionId: "native"})).toEqual({text: "Available shared resource"});
+  expect(runHandoffAction).toHaveBeenCalledWith(expect.any(URL), "list", dir, undefined);
+  expect(hooks.get("before_prompt_build")({}, {sessionId: "native"})).toBeUndefined();
+});
+it.each([
+  ["empty transcript", ""],
+  ["non-session header", JSON.stringify({type: "message", cwd: "/tmp"})],
+  ["missing cwd", JSON.stringify({type: "session"})],
+  ["relative cwd", JSON.stringify({type: "session", cwd: "relative/project"})],
+])("rejects %s before accessing shared resources", async (_description, content) => {
+  const {cmd, hooks} = setup();
+  const sessionFile = join(dir, "invalid.jsonl");
+  await writeFile(sessionFile, content);
+  const result = await cmd.handler({args: "resume /tmp/shared.json", sessionFile, sessionId: "native"});
+  expect(result.text).toContain("project directory is unavailable");
+  expect(runHandoffAction).not.toHaveBeenCalled();
+  expect(hooks.get("before_prompt_build")({}, {sessionId: "native"})).toBeUndefined();
+});
+it("reports a missing transcript rather than guessing a project", async () => {
+  const {cmd} = setup();
+  expect((await cmd.handler({args: "list", sessionFile: join(dir, "missing.jsonl")})).text).toContain("ENOENT");
+  expect(runHandoffAction).not.toHaveBeenCalled();
+});
+it("rejects tools without a native workspace and unsupported actions", async () => {
+  const missing = setup({}).tool;
+  expect(await missing.execute("call", {action: "list"})).toMatchObject({isError: true, content: [{text: expect.stringContaining("project directory is unavailable")}]});
+  const {tool} = setup();
+  expect(await tool.execute("call", {action: "save"})).toMatchObject({isError: true, content: [{text: expect.stringContaining("Choose list or resume")}]});
+  expect(runHandoffAction).not.toHaveBeenCalled();
+});
+it("lists through the tool without requiring a resource path", async () => {
+  const {tool} = setup();
+  vi.mocked(runHandoffAction).mockResolvedValue("No shared resources");
+  expect(await tool.execute("call", {action: "list"})).toEqual({content: [{type: "text", text: "No shared resources"}]});
+  expect(runHandoffAction).toHaveBeenCalledWith(expect.any(URL), "list", "/tmp/native-project", undefined);
+});
+it.each([new Error("Resource cannot be read"), "Resource cannot be read"])("surfaces resource failures as tool errors: %s", async (error) => {
+  const {tool} = setup();
+  vi.mocked(runHandoffAction).mockRejectedValue(error);
+  expect(await tool.execute("call", {action: "resume", resource: "/tmp/shared.json"})).toEqual({isError: true, content: [{type: "text", text: "Session handoff failed: Resource cannot be read"}]});
+});
+it("reports a non-Error command failure without injecting context", async () => {
+  const {cmd, hooks} = setup();
+  const sessionFile = join(dir, "native.jsonl");
+  await writeFile(sessionFile, JSON.stringify({type: "session", cwd: dir}) + "\n");
+  vi.mocked(runHandoffAction).mockRejectedValue("Resource cannot be read");
+  expect(await cmd.handler({args: "resume /tmp/shared.json", sessionFile, sessionId: "native"})).toEqual({text: "Session handoff failed: Resource cannot be read"});
+  expect(hooks.get("before_prompt_build")({}, {sessionId: "native"})).toBeUndefined();
+});