Files
pikasTech-HWLAB/internal/cloud/server-agent-chat.test.ts
T
2026-06-01 17:06:11 +08:00

1042 lines
42 KiB
TypeScript

import assert from "node:assert/strict";
import { createServer as createHttpServer } from "node:http";
import { mkdtemp, mkdir, rm, writeFile } from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { test } from "bun:test";
import { createCloudApiServer } from "./server.ts";
import { validateCodeAgentChatSchema } from "./code-agent-chat.ts";
import { createCodexStdioSessionManager } from "./codex-stdio-session.ts";
import { createCodeAgentTraceStore } from "./code-agent-trace-store.ts";
import {
codexStdioChatFixture,
codexStdioReadyFixture,
createFakeAppServerClient,
createFakeCodexCommand,
delay,
pollAgentResult,
postAgent,
postAgentRaw,
prepareFakeCodexHome
} from "./server-test-helpers.ts";
test("cloud api /v1/agent/chat parse and params errors use structured blocker envelope", async () => {
const server = createCloudApiServer({
env: {
HWLAB_CODE_AGENT_MODEL: "gpt-test"
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const parse = await postAgentRaw(port, "{", { "x-trace-id": "trc_parse_error" });
assert.equal(parse.status, 400);
assert.equal(parse.body.status, "failed");
assert.equal(parse.body.error.code, "parse_error");
assert.equal(parse.body.error.layer, "api");
assert.equal(parse.body.error.retryable, true);
assert.equal(parse.body.error.traceId, "trc_parse_error");
assert.equal(parse.body.error.route, "/v1/agent/chat");
assert.match(parse.body.error.userMessage, /JSON/u);
assert.equal(Object.hasOwn(parse.body, "reply"), false);
assert.equal(JSON.stringify(parse.body).includes("sk-"), false);
const invalid = await postAgentRaw(port, JSON.stringify([]), { "x-trace-id": "trc_invalid_params" });
assert.equal(invalid.status, 400);
assert.equal(invalid.body.error.code, "invalid_params");
assert.equal(invalid.body.error.layer, "api");
assert.equal(invalid.body.error.retryable, true);
assert.equal(invalid.body.error.blocker.code, "invalid_params");
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat supports short submit and result polling", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-short-"));
const codexHome = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-short-codex-home-"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_CODEX_SANDBOX: "danger-full-access",
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
OPENAI_API_KEY: "test-openai-key-material",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
codexStdioManager: {
describe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access"
};
},
async probe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access"
};
},
async chat(params = {}) {
return {
...codexStdioChatFixture({ workspace, codexHome, params }),
sandbox: "danger-full-access",
session: {
...codexStdioChatFixture({ workspace, codexHome, params }).session,
sandbox: "danger-full-access"
}
};
},
cancel() {},
reapIdle() {}
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const traceId = "trc_server-test-short-submit";
const submit = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": traceId,
"prefer": "respond-async",
"x-hwlab-short-connection": "1"
},
body: JSON.stringify({
conversationId: "cnv_server-test-short-submit",
message: "用pwd列出你当前的工作目录"
})
});
assert.equal(submit.status, 202);
const accepted = await submit.json();
assert.equal(accepted.accepted, true);
assert.equal(accepted.shortConnection, true);
assert.equal(accepted.traceId, traceId);
assert.equal(accepted.resultUrl, `/v1/agent/chat/result/${traceId}`);
const payload = await pollAgentResult(port, traceId);
validateCodeAgentChatSchema(payload);
assert.equal(payload.status, "completed");
assert.equal(payload.traceId, traceId);
assert.equal(payload.sandbox, "danger-full-access");
assert.match(payload.reply.content, new RegExp(workspace.replace(/[.*+?^${}()|[\]\\]/gu, "\\$&")));
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(codexHome, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat delegates v0.2 turns to AgentRun v0.1 over adapter", async () => {
const calls = [];
const agentRunServer = createHttpServer(async (request, response) => {
const url = new URL(request.url || "/", "http://127.0.0.1");
const chunks = [];
for await (const chunk of request) chunks.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk));
const body = chunks.length ? JSON.parse(Buffer.concat(chunks).toString("utf8")) : null;
calls.push({ method: request.method, path: url.pathname, search: url.search, body });
const send = (data) => {
response.writeHead(200, { "content-type": "application/json" });
response.end(`${JSON.stringify({ ok: true, data, traceId: "trc_fake_agentrun" })}\n`);
};
if (request.method === "POST" && url.pathname === "/api/v1/runs") {
assert.equal(body.tenantId, "hwlab");
assert.equal(body.backendProfile, "deepseek");
assert.equal(body.resourceBundleRef.repoUrl, "http://git-mirror-http.devops-infra.svc.cluster.local/pikasTech/HWLAB.git");
assert.equal(body.resourceBundleRef.commitId, "0123456789abcdef0123456789abcdef01234567");
return send({ id: "run_hwlab_adapter", status: "pending", backendProfile: "deepseek", sessionRef: body.sessionRef, resourceBundleRef: body.resourceBundleRef });
}
if (request.method === "POST" && url.pathname === "/api/v1/runs/run_hwlab_adapter/commands") {
assert.equal(body.type, "turn");
assert.match(body.payload.prompt, /AgentRun adapter/u);
return send({ id: "cmd_hwlab_adapter", runId: "run_hwlab_adapter", state: "pending", type: "turn", seq: 1 });
}
if (request.method === "POST" && url.pathname === "/api/v1/runs/run_hwlab_adapter/runner-jobs") {
assert.equal(body.commandId, "cmd_hwlab_adapter");
return send({
action: "create-kubernetes-job",
runId: "run_hwlab_adapter",
commandId: "cmd_hwlab_adapter",
attemptId: "attempt_hwlab_adapter",
runnerId: "runner_hwlab_adapter",
namespace: "agentrun-v01",
jobName: "agentrun-v01-runner-hwlab-adapter",
jobIdentity: { namespace: "agentrun-v01", name: "agentrun-v01-runner-hwlab-adapter" },
runner: { attemptId: "attempt_hwlab_adapter", runnerId: "runner_hwlab_adapter" }
});
}
if (request.method === "GET" && url.pathname === "/api/v1/runs/run_hwlab_adapter/events") {
return send({ items: [
{ id: "evt_1", runId: "run_hwlab_adapter", seq: 1, type: "backend_status", payload: { phase: "runner-job-created", commandId: "cmd_hwlab_adapter", attemptId: "attempt_hwlab_adapter", jobName: "agentrun-v01-runner-hwlab-adapter", namespace: "agentrun-v01" }, createdAt: "2026-06-01T00:00:00.000Z" },
{ id: "evt_2", runId: "run_hwlab_adapter", seq: 2, type: "assistant_message", payload: { text: "AgentRun adapter 已接管 HWLAB Code Agent。" }, createdAt: "2026-06-01T00:00:01.000Z" },
{ id: "evt_3", runId: "run_hwlab_adapter", seq: 3, type: "terminal_status", payload: { terminalStatus: "completed" }, createdAt: "2026-06-01T00:00:02.000Z" }
] });
}
if (request.method === "GET" && url.pathname === "/api/v1/runs/run_hwlab_adapter/commands/cmd_hwlab_adapter/result") {
return send({
runId: "run_hwlab_adapter",
commandId: "cmd_hwlab_adapter",
attemptId: "attempt_hwlab_adapter",
runnerId: "runner_hwlab_adapter",
jobName: "agentrun-v01-runner-hwlab-adapter",
namespace: "agentrun-v01",
status: "completed",
runStatus: "completed",
commandState: "completed",
terminalStatus: "completed",
completed: true,
reply: "AgentRun adapter 已接管 HWLAB Code Agent。",
lastSeq: 3,
eventCount: 3
});
}
response.writeHead(404, { "content-type": "application/json" });
response.end(`${JSON.stringify({ ok: false, failureKind: "schema-invalid", message: `unexpected ${request.method} ${url.pathname}`, traceId: "trc_fake_agentrun" })}\n`);
});
await new Promise((resolve) => agentRunServer.listen(0, "127.0.0.1", resolve));
const agentRunPort = agentRunServer.address().port;
const traceStore = createCodeAgentTraceStore();
const server = createCloudApiServer({
traceStore,
env: {
HWLAB_CODE_AGENT_ADAPTER: "agentrun-v01",
AGENTRUN_MGR_URL: `http://127.0.0.1:${agentRunPort}`,
HWLAB_CODE_AGENT_AGENTRUN_ALLOW_NON_K3S_URL: "1",
HWLAB_CODE_AGENT_AGENTRUN_SOURCE_COMMIT: "0123456789abcdef0123456789abcdef01234567",
HWLAB_CODE_AGENT_DEFAULT_PROVIDER_PROFILE: "deepseek",
HWLAB_CODE_AGENT_DEEPSEEK_MODEL: "deepseek-chat"
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const traceId = "trc_server-test-agentrun-adapter";
const submit = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: { "content-type": "application/json", "x-trace-id": traceId },
body: JSON.stringify({ conversationId: "cnv_server-test-agentrun", message: "AgentRun adapter smoke" })
});
assert.equal(submit.status, 202);
const accepted = await submit.json();
assert.equal(accepted.shortConnection, true);
const payload = await pollAgentResult(port, traceId);
validateCodeAgentChatSchema(payload);
assert.equal(payload.status, "completed");
assert.equal(payload.provider, "agentrun-v01");
assert.equal(payload.backend, "agentrun-v01/deepseek");
assert.equal(payload.agentRun.runId, "run_hwlab_adapter");
assert.equal(payload.agentRun.commandId, "cmd_hwlab_adapter");
assert.equal(payload.agentRun.jobName, "agentrun-v01-runner-hwlab-adapter");
assert.match(payload.reply.content, /AgentRun adapter/u);
assert.ok(payload.runnerTrace.events.some((event) => event.label === "agentrun:backend:runner-job-created"));
assert.ok(calls.some((call) => call.path === "/api/v1/runs/run_hwlab_adapter/runner-jobs"));
const trace = await fetch(`http://127.0.0.1:${port}/v1/agent/chat/trace/${traceId}`);
assert.equal(trace.status, 200);
const traceBody = await trace.json();
assert.ok(traceBody.events.some((event) => event.runId === "run_hwlab_adapter"));
assert.ok(traceBody.events.some((event) => event.label === "agentrun:assistant:message"));
} finally {
await new Promise((resolve, reject) => server.close((error) => (error ? reject(error) : resolve())));
await new Promise((resolve, reject) => agentRunServer.close((error) => (error ? reject(error) : resolve())));
}
});
test("cloud api AgentRun adapter rejects non-internal manager URLs by default", async () => {
const server = createCloudApiServer({
env: {
HWLAB_CODE_AGENT_ADAPTER: "agentrun-v01",
AGENTRUN_MGR_URL: "http://74.48.78.17:8080"
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: { "content-type": "application/json", "x-trace-id": "trc_server-test-agentrun-public-url" },
body: JSON.stringify({ conversationId: "cnv_server-test-agentrun-public-url", message: "must reject public manager url" })
});
assert.equal(response.status, 500);
const body = await response.json();
assert.equal(body.error.code, "internal_error");
assert.match(body.error.reason, /internal k3s Service DNS/u);
} finally {
await new Promise((resolve, reject) => server.close((error) => (error ? reject(error) : resolve())));
}
});
test("cloud api /v1/agent/chat records authenticated owner on agent session", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-owner-"));
const codexHome = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-owner-codex-home-"));
const ownerRecords = [];
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_CODEX_SANDBOX: "danger-full-access",
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
OPENAI_API_KEY: "test-openai-key-material",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
accessController: {
required: true,
async authenticate() {
return { ok: true, actor: { id: "usr_agent_owner", username: "agent-owner", displayName: "Agent Owner", role: "user", status: "active" } };
},
async recordAgentSessionOwner(input) {
ownerRecords.push(input);
return input;
}
},
codexStdioManager: {
describe() { return { ...codexStdioReadyFixture({ workspace, codexHome }), sandbox: "danger-full-access" }; },
async probe() { return { ...codexStdioReadyFixture({ workspace, codexHome }), sandbox: "danger-full-access" }; },
async chat(params = {}) {
return {
...codexStdioChatFixture({ workspace, codexHome, params }),
sandbox: "danger-full-access"
};
},
cancel() {},
reapIdle() {}
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const payload = await postAgent(port, {
conversationId: "cnv_server-test-owner",
traceId: "trc_server-test-owner",
message: "owner binding smoke"
});
validateCodeAgentChatSchema(payload);
assert.equal(payload.ownerUserId, "usr_agent_owner");
assert.equal(ownerRecords.length, 1);
assert.equal(ownerRecords[0].ownerUserId, "usr_agent_owner");
assert.equal(ownerRecords[0].sessionId, "ses_server_stdio_pwd");
assert.equal(ownerRecords[0].conversationId, "cnv_server-test-owner");
assert.equal(ownerRecords[0].traceId, "trc_server-test-owner");
assert.equal(ownerRecords[0].session.valuesRedacted, true);
assert.equal(JSON.stringify(ownerRecords).includes("test-openai-key-material"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(codexHome, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat/inspect maps conversation session and thread details to trace evidence", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-inspect-"));
const codexHome = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-inspect-codex-home-"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_CODEX_SANDBOX: "danger-full-access",
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
OPENAI_API_KEY: "test-openai-key-material",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
codexStdioManager: {
describe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access",
recentSessions: [{
sessionId: "ses_server_stdio_pwd",
conversationId: "cnv_server-test-inspect",
threadId: "019e6c72-2373-75e3-9856-eff286db7163",
status: "idle",
currentTraceId: null,
lastTraceId: "trc_server-test-inspect",
workspace,
sandbox: "danger-full-access",
secretMaterialStored: false,
valuesRedacted: true
}]
};
},
async probe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access"
};
},
async chat(params = {}) {
const base = codexStdioChatFixture({ workspace, codexHome, params });
return {
...base,
sandbox: "danger-full-access",
session: {
...base.session,
threadId: "019e6c72-2373-75e3-9856-eff286db7163",
sandbox: "danger-full-access"
},
sessionReuse: {
...base.sessionReuse,
threadId: "019e6c72-2373-75e3-9856-eff286db7163"
}
};
},
get(sessionId) {
if (sessionId !== "ses_server_stdio_pwd") return null;
return {
sessionId,
conversationId: "cnv_server-test-inspect",
threadId: "019e6c72-2373-75e3-9856-eff286db7163",
status: "idle",
lastTraceId: "trc_server-test-inspect",
workspace,
sandbox: "danger-full-access",
secretMaterialStored: false,
valuesRedacted: true
};
},
cancel() {},
reapIdle() {}
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const traceId = "trc_server-test-inspect";
const submit = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": traceId,
"prefer": "respond-async",
"x-hwlab-short-connection": "1"
},
body: JSON.stringify({
conversationId: "cnv_server-test-inspect",
message: "inspect mapping smoke"
})
});
assert.equal(submit.status, 202);
const payload = await pollAgentResult(port, traceId);
assert.equal(payload.status, "completed");
const inspect = await fetch(`http://127.0.0.1:${port}/v1/agent/chat/inspect?conversationId=cnv_server-test-inspect&sessionId=ses_server_stdio_pwd&threadId=019e6c72-2373-75e3-9856-eff286db7163`);
assert.equal(inspect.status, 200);
const body = await inspect.json();
assert.equal(body.ok, true);
assert.equal(body.status, "found");
assert.equal(body.latestTraceId, traceId);
assert.equal(body.traceUrl, `/v1/agent/chat/trace/${traceId}`);
assert.equal(body.resultUrl, `/v1/agent/chat/result/${traceId}`);
assert.equal(body.query.threadId, "019e6c72-2373-75e3-9856-eff286db7163");
assert.equal(body.session.sessionId, "ses_server_stdio_pwd");
assert.equal(body.conversationFacts.conversationId, "cnv_server-test-inspect");
assert.equal(body.runnerTrace.traceId, traceId);
assert.equal(body.valuesRedacted, true);
assert.equal(body.secretMaterialStored, false);
assert.equal(JSON.stringify(body).includes("test-openai-key-material"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(codexHome, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat/inspect returns not_found for empty conversation facts", async () => {
const server = createCloudApiServer();
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat/inspect?conversationId=cnv_server-test-inspect-missing&sessionId=ses_server_test_missing&threadId=019e6c72-2373-75e3-9856-eff286db7163&traceId=trc_server_test_missing`);
assert.equal(response.status, 404);
const body = await response.json();
assert.equal(body.ok, false);
assert.equal(body.status, "not_found");
assert.equal(body.latestTraceId, null);
assert.deepEqual(body.traceIds, []);
assert.equal(body.traceUrl, null);
assert.equal(body.resultUrl, null);
assert.equal(body.conversationFacts.turnCount, 0);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api result polling compacts large runnerTrace while preserving providerTrace", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-result-compact-"));
const codexHome = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-result-compact-codex-home-"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_CODEX_SANDBOX: "danger-full-access",
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
HWLAB_CODE_AGENT_RESULT_TRACE_EVENT_LIMIT: "32",
OPENAI_API_KEY: "test-openai-key-material",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
codexStdioManager: {
describe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access"
};
},
async probe() {
return {
...codexStdioReadyFixture({ workspace, codexHome }),
sandbox: "danger-full-access"
};
},
async chat(params = {}) {
const base = codexStdioChatFixture({ workspace, codexHome, params });
const events = Array.from({ length: 160 }, (_, index) => ({
seq: index + 1,
traceId: params.traceId,
type: "assistant_message",
status: "chunk",
label: index === 159 ? "assistant:completed" : "assistant:chunk",
createdAt: "2026-05-23T00:00:00.000Z",
waitingFor: index === 159 ? null : "turn/completed",
valuesPrinted: false
}));
return {
...base,
sandbox: "danger-full-access",
session: {
...base.session,
sandbox: "danger-full-access"
},
runnerTrace: {
...base.runnerTrace,
events,
eventLabels: events.map((event) => event.label),
eventCount: events.length,
lastEvent: events.at(-1)
}
};
},
cancel() {},
reapIdle() {}
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const traceId = "trc_server-test-result-compact";
const submit = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": traceId,
"prefer": "respond-async",
"x-hwlab-short-connection": "1"
},
body: JSON.stringify({
conversationId: "cnv_server-test-result-compact",
message: "生成大量 trace 后返回"
})
});
assert.equal(submit.status, 202);
const payload = await pollAgentResult(port, traceId);
validateCodeAgentChatSchema(payload);
assert.equal(payload.status, "completed");
assert.equal(payload.traceId, traceId);
assert.equal(payload.providerTrace.protocol, "codex-app-server-jsonrpc-stdio");
assert.equal(payload.providerTrace.command, "codex app-server --listen stdio://");
assert.equal(payload.runnerTrace.eventCount, 160);
assert.equal(payload.runnerTrace.eventsCompacted, true);
assert.equal(payload.runnerTrace.events.length, 32);
assert.equal(payload.runnerTrace.events.some((event) => event.label === "trace:compacted"), true);
assert.equal(payload.runnerTrace.lastEvent.label, "assistant:completed");
assert.equal(payload.runnerTrace.eventLabels.includes("trace:compacted"), true);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(codexHome, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat answers skills prompt through real Codex stdio and retains trace", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-stdio-skills-"));
const codexHome = path.join(workspace, "codex-home");
const skillsDir = path.join(workspace, "skills");
const fakeCodex = await createFakeCodexCommand();
await mkdir(path.join(skillsDir, "hwlab-test-skill"), { recursive: true });
await prepareFakeCodexHome(codexHome);
await writeFile(path.join(skillsDir, "hwlab-test-skill", "SKILL.md"), [
"---",
"name: hwlab-test-skill",
"description: Test skill mounted for Codex stdio discovery.",
"version: v-test",
"---",
"",
"# HWLAB Test Skill"
].join("\n"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
OPENAI_API_KEY: "test-openai-key-material",
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
HWLAB_CODE_AGENT_CODEX_COMMAND: fakeCodex.command,
HWLAB_CODE_AGENT_CODEX_STDIO_ENABLED: "1",
HWLAB_CODE_AGENT_CODEX_STDIO_SUPERVISOR: "repo-owned",
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_SKILLS_DIRS: skillsDir,
HWLAB_CODE_AGENT_SKILLS_STRICT: "1",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
codexStdioManager: createCodexStdioSessionManager({
idFactory: () => "ses_server_stdio_skills",
createRpcClient: async () => createFakeAppServerClient({
text: "模型泛化回答不应成为 skills 最终回复。"
})
})
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const payload = await postAgent(port, {
conversationId: "cnv_server_stdio_skills",
traceId: "trc_server_stdio_skills",
message: "列出你可用的skills"
});
validateCodeAgentChatSchema(payload);
assert.equal(payload.status, "completed");
assert.equal(payload.provider, "codex-stdio");
assert.equal(payload.backend, "hwlab-cloud-api/codex-app-server-stdio");
assert.equal(payload.capabilityLevel, "long-lived-codex-stdio-session");
assert.equal(payload.providerTrace.sidecarOnly, undefined);
assert.equal(payload.providerTrace.toolName, "codex-app-server.thread/start+turn/start");
assert.equal(payload.skills.status, "ready");
const discovered = payload.skills.items.find((skill) => skill.name === "hwlab-test-skill");
assert.ok(discovered);
assert.equal(discovered.source, path.join(skillsDir, "hwlab-test-skill", "SKILL.md"));
assert.equal(discovered.manifest.path, path.join(skillsDir, "hwlab-test-skill", "SKILL.md"));
assert.equal(discovered.version, "v-test");
assert.ok(payload.toolCalls.some((toolCall) => toolCall.name === "skills.discover" && toolCall.status === "completed"));
assert.match(payload.reply.content, /hwlab-test-skill/u);
assert.match(payload.reply.content, /真实 skill manifest 清单/u);
assert.doesNotMatch(payload.reply.content, /模型泛化回答/u);
assert.ok(payload.runnerTrace.events.some((event) => event.label === "prompt:sent"));
assert.ok(payload.runnerTrace.events.some((event) => event.label === "tool:codex-app-server.thread/start+turn/start:started"));
assert.equal(JSON.stringify(payload).includes("test-openai-key-material"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(fakeCodex.root, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat returns structured skills blocker without local skills manifest", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-stdio-skills-missing-"));
const codexHome = path.join(workspace, "codex-home");
const fakeCodex = await createFakeCodexCommand();
await prepareFakeCodexHome(codexHome);
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
OPENAI_API_KEY: "test-openai-key-material",
CODEX_HOME: codexHome,
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
HWLAB_CODE_AGENT_CODEX_COMMAND: fakeCodex.command,
HWLAB_CODE_AGENT_CODEX_STDIO_ENABLED: "1",
HWLAB_CODE_AGENT_CODEX_STDIO_SUPERVISOR: "repo-owned",
HWLAB_CODE_AGENT_CODEX_WORKSPACE: workspace,
HWLAB_CODE_AGENT_SKILLS_DIRS: path.join(workspace, "missing-skills"),
HWLAB_CODE_AGENT_SKILLS_STRICT: "1",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: "http://127.0.0.1:65535/v1/responses"
},
codexStdioManager: createCodexStdioSessionManager({
idFactory: () => "ses_server_stdio_skills_missing",
createRpcClient: async () => createFakeAppServerClient({
text: "模型泛化回答:alpha-skill"
})
})
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const payload = await postAgent(port, {
conversationId: "cnv_server_stdio_skills_missing",
traceId: "trc_server_stdio_skills_missing",
message: "列出你可用的skills"
});
validateCodeAgentChatSchema(payload);
assert.equal(payload.status, "failed");
assert.equal(payload.provider, "codex-stdio");
assert.equal(payload.backend, "hwlab-cloud-api/codex-app-server-stdio");
assert.equal(payload.capabilityLevel, "blocked");
assert.equal(payload.error.code, "skills_unavailable");
assert.equal(payload.skills.status, "blocked");
assert.equal(payload.skills.blockers[0].code, "skills_unavailable");
assert.deepEqual(payload.skills.sourceIssues, ["pikasTech/HWLAB#136", "pikasTech/HWLAB#237"]);
assert.ok(payload.toolCalls.some((toolCall) => toolCall.name === "skills.discover" && toolCall.status === "blocked"));
assert.equal(payload.toolCalls.some((toolCall) => toolCall.name === "codex-app-server.thread/start+turn/start"), false);
assert.equal(payload.session.status, "idle");
assert.equal(Object.hasOwn(payload, "reply"), false);
assert.equal(JSON.stringify(payload).includes("alpha-skill"), false);
assert.equal(JSON.stringify(payload).includes("test-openai-key-material"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await rm(workspace, { recursive: true, force: true });
await rm(fakeCodex.root, { recursive: true, force: true });
}
});
test("cloud api /v1/agent/chat blocks forbidden file paths without leaking values", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-file-block-"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
HWLAB_CODE_AGENT_WORKSPACE: workspace
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const payload = await postAgent(port, {
conversationId: "cnv_server-test-security-block",
traceId: "trc_server-test-security-block",
message: "cat ../outside.txt"
});
assert.equal(payload.status, "failed");
assert.match(payload.error.code, /^(codex_cli_binary_missing|stdio_protocol_not_wired)$/u);
assert.deepEqual(payload.toolCalls, []);
assert.equal(payload.capabilityLevel, "blocked");
assert.ok(payload.runnerTrace.events.some((event) => event.label === "codex-stdio:blocked"));
assert.equal(Object.hasOwn(payload, "reply"), false);
assert.equal(JSON.stringify(payload).includes("sk-"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat/cancel reports unsupported lifecycle degradation", async () => {
const traceStore = createCodeAgentTraceStore();
traceStore.append("trc_cancel_unsupported", {
type: "request",
status: "running",
label: "request:running",
sessionId: "ses_cancel_unsupported",
sessionStatus: "busy",
sessionLifecycleStatus: "busy"
});
const server = createCloudApiServer({
traceStore,
codexStdioManager: {
get() {
return null;
}
},
env: {
PATH: process.env.PATH,
HWLAB_CODE_AGENT_PROVIDER: "codex-stdio"
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat/cancel`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": "trc_cancel_unsupported"
},
body: JSON.stringify({
conversationId: "cnv_cancel_unsupported",
sessionId: "ses_cancel_unsupported",
traceId: "trc_cancel_unsupported"
})
});
assert.equal(response.status, 501);
const payload = await response.json();
assert.equal(payload.status, "degraded");
assert.equal(payload.unsupported, true);
assert.equal(payload.degraded, true);
assert.equal(payload.error.code, "cancel_unsupported");
assert.equal(payload.sessionLifecycleStatus, "failed");
assert.equal(payload.sessionSummary.unsupported, true);
assert.match(payload.sessionSummary.userMessage, /unsupported\/degraded/u);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat reports Codex stdio blocker instead of structured skills fallback", async () => {
const root = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-no-skills-"));
const skillsDir = path.join(root, "missing-skills");
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
HWLAB_CODE_AGENT_WORKSPACE: root
},
skillsDirs: [skillsDir],
skillsDirsExact: true
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json"
},
body: JSON.stringify({
conversationId: "cnv_server-test-skills-missing",
message: "列出你可用的skills"
})
});
assert.equal(response.status, 200);
const payload = await response.json();
assert.equal(payload.status, "failed");
assert.equal(payload.provider, "codex-stdio");
assert.notEqual(payload.error.code, "skills_unavailable");
assert.equal(payload.capabilityLevel, "blocked");
assert.equal(payload.skills.status, "not_requested");
assert.deepEqual(payload.toolCalls, []);
assert.ok(payload.runnerLimitations.includes("no-controlled-readonly-fallback"));
assert.equal(Object.hasOwn(payload, "reply"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat does not run unsupported local grep fallback", async () => {
const workspace = await mkdtemp(path.join(os.tmpdir(), "hwlab-agent-tool-blocked-"));
const server = createCloudApiServer({
env: {
PATH: process.env.PATH,
HWLAB_CODE_AGENT_WORKSPACE: workspace
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json"
},
body: JSON.stringify({
conversationId: "cnv_server-test-tool-unavailable",
message: "请用grep搜索 package.json"
})
});
assert.equal(response.status, 200);
const payload = await response.json();
assert.equal(payload.status, "failed");
assert.notEqual(payload.error.code, "tool_unavailable");
assert.equal(payload.provider, "codex-stdio");
assert.equal(payload.capabilityLevel, "blocked");
assert.deepEqual(payload.toolCalls, []);
assert.ok(payload.runnerLimitations.includes("no-controlled-readonly-fallback"));
assert.equal(Object.hasOwn(payload, "reply"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat does not call delayed provider fallback beyond legacy 4500ms UI timeout", async () => {
let providerCalled = false;
const server = createCloudApiServer({
callCodeAgentProvider: async ({ providerPlan, message, traceId }) => {
providerCalled = true;
await delay(4700);
return {
provider: providerPlan.provider,
model: providerPlan.model,
backend: providerPlan.backend,
content: `延迟真实 provider stub: ${message} / ${traceId}`,
usage: null,
providerTrace: {
source: "delayed-test-provider"
}
};
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const startedAt = Date.now();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": "trc_server-test-agent-chat-delayed"
},
body: JSON.stringify({
conversationId: "cnv_server-test-agent-chat-delayed",
message: "请用一句话说明当前 HWLAB 工作台可以做什么,稍后回答"
})
});
assert.equal(response.status, 200);
const payload = await response.json();
assert.equal(Date.now() - startedAt < 4500, true);
assert.equal(payload.status, "failed");
assert.equal(payload.conversationId, "cnv_server-test-agent-chat-delayed");
assert.equal(payload.traceId, "trc_server-test-agent-chat-delayed");
assert.equal(payload.provider, "codex-stdio");
assert.equal(payload.capabilityLevel, "blocked");
assert.equal(providerCalled, false);
assert.equal(Object.hasOwn(payload, "reply"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});
test("cloud api /v1/agent/chat does not call OpenAI provider 502/503 fallback", async () => {
for (const status of [502, 503]) {
const providerServer = createHttpServer((request, response) => {
request.resume();
response.writeHead(status, { "content-type": "application/json" });
response.end(JSON.stringify({
error: {
message: `upstream ${status}`
}
}));
});
await new Promise((resolve) => providerServer.listen(0, "127.0.0.1", resolve));
const providerPort = providerServer.address().port;
const server = createCloudApiServer({
env: {
OPENAI_API_KEY: "test-openai-key-material",
HWLAB_CODE_AGENT_PROVIDER: "openai",
HWLAB_CODE_AGENT_ALLOW_TEXT_FALLBACK: "true",
HWLAB_CODE_AGENT_MODEL: "gpt-test",
HWLAB_CODE_AGENT_OPENAI_BASE_URL: `http://127.0.0.1:${providerPort}/v1/responses`
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json",
"x-trace-id": `trc_server-test-agent-chat-provider-${status}`
},
body: JSON.stringify({
conversationId: `cnv_server-test-agent-chat-provider-${status}`,
message: `provider ${status}`
})
});
assert.equal(response.status, 200);
const payload = await response.json();
assert.equal(payload.status, "failed");
assert.equal(payload.traceId, `trc_server-test-agent-chat-provider-${status}`);
assert.match(payload.error.code, /^(codex_cli_binary_missing|stdio_protocol_not_wired)$/u);
assert.match(payload.error.layer, /^(runner|api)$/u);
assert.equal(typeof payload.error.retryable, "boolean");
assert.equal(payload.error.blocker.traceId, `trc_server-test-agent-chat-provider-${status}`);
assert.equal(typeof payload.error.blocker.retryable, "boolean");
assert.equal(payload.error.blocker.capabilityLevel, "blocked");
assert.equal(payload.provider, "codex-stdio");
assert.equal(Object.hasOwn(payload, "reply"), false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
await new Promise((resolve, reject) => {
providerServer.close((error) => (error ? reject(error) : resolve()));
});
}
}
});
test("cloud api /v1/agent/chat does not call empty provider text fallback", async () => {
let providerCalled = false;
const server = createCloudApiServer({
callCodeAgentProvider: async () => {
providerCalled = true;
throw new Error("empty provider fallback must not be used");
}
});
await new Promise((resolve) => server.listen(0, "127.0.0.1", resolve));
try {
const { port } = server.address();
const response = await fetch(`http://127.0.0.1:${port}/v1/agent/chat`, {
method: "POST",
headers: {
"content-type": "application/json"
},
body: JSON.stringify({
conversationId: "cnv_empty-provider-text",
message: "你好"
})
});
assert.equal(response.status, 200);
const payload = await response.json();
assert.equal(payload.conversationId, "cnv_empty-provider-text");
assert.equal(payload.status, "failed");
assert.match(payload.error.code, /^(codex_cli_binary_missing|stdio_protocol_not_wired)$/u);
assert.equal(payload.provider, "codex-stdio");
assert.equal(Object.hasOwn(payload, "reply"), false);
assert.equal(providerCalled, false);
} finally {
await new Promise((resolve, reject) => {
server.close((error) => (error ? reject(error) : resolve()));
});
}
});