Files
my-pi/pi-permission-system/test/handlers/tool-call.test.ts
T
云服务部-叶林立andDevin <158243242+devin-ai-integration[bot]@users.noreply.github.com> 7e6a8faae1 feat(pi-permission-system): AutoReview 改为跟随主模型的对话式工具调用审查
AutoReview:

- 审查器回到 Codex Guardian 风格的对话式审查,默认跟随会话当前主模型
  (每次审查实时读取 ctx.model);provider/model 必须同时设置才固定审查
  模型,只设其一视为配置错误并转人工。
- 结论只通过 approve / reject 两个工具给出,参数仅 reason;不强制
  toolChoice,正文文本(即使是合法 JSON)一律忽略。
- approve 自动放行;reject 转人工并在弹窗显示理由(标注模型生成)。
  超时、Provider 错误、截断、未调用工具、多次调用、参数无效等异常同样
  转人工,并附固定的系统说明;用户按 Esc 中止时直接拒绝。
- 模型调用改为 modelRegistry.streamSimple(...).result():Pi 1.3.0 面向
  扩展的 ModelRegistry 没有 completeSimple,原写法在真实环境中每次都会
  转人工。
- 对话记录保留可信度标注;工具结果在有更短的 pi-compaction 摘要时替换
  为摘要;token 估算改为非 ASCII 字符按 1 token 计;预算可配置。
- 同一条用户消息内完全相同的请求只缓存 approve 结论。
- 默认 reasoning low、总超时 30 秒、最多 2 次尝试、maxOutputTokens 4000。

删除:

- 未发布过的 TypeSafe Jev 分类器实现、DenialCircuitBreaker。
- 整个 annotator 机制(model-explanation、注册表、registerAnnotator、
  provider-timeout);annotators 配置键保留为已弃用并忽略,避免旧配置
  被拒绝。

测试:

- 新增 reviewer / transcript / 弹窗说明 / 默认配置测试;清理重复、恒真
  或只测实现细节的旧测试。
- 使用 deepseek-v4.1-flash 实测 5 个场景各 5 次,25 次结论全部符合预期。

Generated with [Devin](https://devin.ai)

Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-10-19 01:54:17 -07:00

301 lines
11 KiB
TypeScript

import { describe, expect, it, vi } from "vitest";
import { getEventInput } from "#src/handlers/permission-gate-handler";
import { findEvidence } from "#src/presentation/prompt-payload";
import {
makeBashCommandCheck,
makeCheckResult,
makeCtx,
makeHandler,
makeSurfaceCheck,
makeToolCallEvent,
} from "#test/helpers/handler-fixtures";
// ── SDK stubs ──────────────────────────────────────────────────────────────
vi.mock("@earendil-works/pi-coding-agent", async (importOriginal) => {
const original =
await importOriginal<typeof import("@earendil-works/pi-coding-agent")>();
return { ...original };
});
// ── getEventInput ──────────────────────────────────────────────────────────
describe("getEventInput", () => {
it("returns the input field when present", () => {
expect(getEventInput({ input: { path: "/foo" } })).toEqual({
path: "/foo",
});
});
it("returns the arguments field when input is absent", () => {
expect(getEventInput({ arguments: { command: "ls" } })).toEqual({
command: "ls",
});
});
it("returns empty object when neither field is present", () => {
expect(getEventInput({ type: "tool_call" })).toEqual({});
});
it("prefers input over arguments when both are present", () => {
expect(getEventInput({ input: { a: 1 }, arguments: { b: 2 } })).toEqual({
a: 1,
});
});
});
// ── handleToolCall ─────────────────────────────────────────────────────────
describe("handleToolCall", () => {
it("blocks when tool is not registered", async () => {
const { handler } = makeHandler({ tools: ["read"] });
const result = await handler.handleToolCall(
makeToolCallEvent("unknown-tool"),
makeCtx(),
);
expect(result).toMatchObject({ action: "block" });
});
it("returns empty object when tool is allowed", async () => {
const { handler } = makeHandler();
const result = await handler.handleToolCall(
makeToolCallEvent("read"),
makeCtx(),
);
expect(result).toEqual({ action: "allow" });
});
it("blocks when tool is denied by policy", async () => {
const { handler } = makeHandler({
session: {
checkPermission: vi
.fn()
.mockReturnValue(makeCheckResult({ state: "deny" })),
},
});
const result = await handler.handleToolCall(
makeToolCallEvent("read"),
makeCtx(),
);
expect(result).toMatchObject({ action: "block" });
});
});
// ── skill-read gate ────────────────────────────────────────────────────────
describe("handleToolCall — skill-read gate", () => {
it("blocks a read of a denied skill path", async () => {
const skillEntry = {
name: "librarian",
description: "Research skills",
location: "/skills/librarian/SKILL.md",
state: "deny" as const,
normalizedLocation: "/skills/librarian/SKILL.md",
normalizedBaseDir: "/skills/librarian",
};
const { handler } = makeHandler({
session: {
getActiveSkillEntries: vi.fn().mockReturnValue([skillEntry]),
},
toolRegistry: {
getAll: vi.fn().mockReturnValue([{ toolName: "read" }]),
},
});
const event = {
type: "tool_call",
toolCallId: "tc-skill",
toolName: "read",
input: { path: "/skills/librarian/SKILL.md" },
};
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
it("allows a read of a non-skill path even when skill entries are present", async () => {
const skillEntry = {
name: "librarian",
description: "Research skills",
location: "/skills/librarian/SKILL.md",
state: "deny" as const,
normalizedLocation: "/skills/librarian/SKILL.md",
normalizedBaseDir: "/skills/librarian",
};
const { handler } = makeHandler({
session: {
getActiveSkillEntries: vi.fn().mockReturnValue([skillEntry]),
},
toolRegistry: {
getAll: vi.fn().mockReturnValue([{ toolName: "read" }]),
},
});
const event = {
type: "tool_call",
toolCallId: "tc-ok",
toolName: "read",
input: { path: "/test/project/src/index.ts" },
};
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toEqual({ action: "allow" });
});
});
// ── external-directory gate ────────────────────────────────────────────────
describe("handleToolCall — external-directory gate", () => {
it("blocks a read of a path outside cwd when policy is deny", async () => {
const { handler } = makeHandler({
session: {
checkPermission: vi
.fn()
.mockReturnValue(makeCheckResult({ state: "deny" })),
},
tools: ["read"],
});
const event = makeToolCallEvent("read", {
input: { path: "/outside/project/file.ts" },
});
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
});
// ── bash external-directory gate ──────────────────────────────────────────
describe("handleToolCall — bash external-directory gate", () => {
it("blocks a bash command referencing an external path when policy is deny", async () => {
const { handler } = makeHandler({
session: {
checkPermission: vi
.fn()
.mockReturnValue(makeCheckResult({ state: "deny" })),
},
tools: ["bash"],
});
const event = makeToolCallEvent("bash", {
input: { command: "cat /outside/project/file.ts" },
});
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
});
// ── path gate (tools) ─────────────────────────────────────────────────────
describe("handleToolCall — path gate (tools)", () => {
it("blocks a read of .env when path surface denies *.env", async () => {
const { handler } = makeHandler({
session: {
checkPermission: makeSurfaceCheck({
path: { state: "deny", matchedPattern: "*.env" },
}),
},
tools: ["read"],
});
const event = makeToolCallEvent("read", { input: { path: ".env" } });
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
it("allows a read when path surface allows", async () => {
const { handler } = makeHandler({ tools: ["read"] });
const event = makeToolCallEvent("read", {
input: { path: "src/index.ts" },
});
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toEqual({ action: "allow" });
});
});
// ── bash path gate ────────────────────────────────────────────────────────
describe("handleToolCall — bash path gate", () => {
it("blocks a bash command accessing .env when path surface denies", async () => {
const { handler } = makeHandler({
session: {
checkPermission: makeSurfaceCheck({
path: { state: "deny", matchedPattern: "*.env" },
}),
},
tools: ["bash"],
});
const event = makeToolCallEvent("bash", { input: { command: "cat .env" } });
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
});
// ── bash command chain gate ───────────────────────────────────────────────
describe("handleToolCall — bash command chain gate", () => {
it("blocks a chain when a later sub-command is denied (#301)", async () => {
const { handler } = makeHandler({
session: {
checkPermission: makeBashCommandCheck({
deny: /^npm\b/,
denyMatched: "npm *",
allowMatched: "echo *",
}),
},
tools: ["bash"],
});
const event = makeToolCallEvent("bash", {
input: { command: "echo start && npm install compromised-package" },
});
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
it("blocks a command nested inside command substitution (#306)", async () => {
const { handler } = makeHandler({
session: {
checkPermission: makeBashCommandCheck({
deny: /^rm\b/,
denyMatched: "rm *",
allowMatched: "echo *",
}),
},
tools: ["bash"],
});
const event = makeToolCallEvent("bash", {
input: { command: "echo $(rm -rf foo)" },
});
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toMatchObject({ action: "block" });
});
it("allows a single non-chained bash command", async () => {
const { handler } = makeHandler({ tools: ["bash"] });
const event = makeToolCallEvent("bash", { input: { command: "echo hi" } });
const result = await handler.handleToolCall(event, makeCtx());
expect(result).toEqual({ action: "allow" });
});
});
describe("handleToolCall — generic ask prompt content", () => {
it("ask prompt includes serialized tool input for informed approval", async () => {
const { handler, prompter } = makeHandler({
session: {
checkPermission: makeSurfaceCheck({
weather_lookup: { state: "ask" },
}),
},
tools: ["weather_lookup"],
prompter: {
escalate: vi
.fn()
.mockResolvedValue({ approved: false, state: "denied" }),
},
});
const event = makeToolCallEvent("weather_lookup", {
input: { city: "Chicago", units: "metric" },
});
await handler.handleToolCall(event, makeCtx());
expect(vi.mocked(prompter.escalate)).toHaveBeenCalledTimes(1);
const promptDetails = vi.mocked(prompter.escalate).mock.calls[0][0];
expect(findEvidence(promptDetails.payload, "input")?.text).toMatch(
/\{"city":"Chicago","units":"metric"\}/,
);
});
});