Files
my-pi/pi-permission-system/test/handlers/gates/tool.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

287 lines
9.8 KiB
TypeScript

import { describe, expect, it } from "vitest";
import type { ShellInvocation } from "#src/access-intent/tool-kind";
import { describeToolGate } from "#src/handlers/gates/tool";
import type { ToolCallContext } from "#src/handlers/gates/types";
import { posixPathFlavor } from "#src/path/path-flavor";
import { PathNormalizer } from "#src/path-normalizer";
import {
TOOL_INPUT_PREVIEW_MAX_LENGTH,
TOOL_TEXT_SUMMARY_MAX_LENGTH,
} from "#src/tool-input-preview";
import { ToolPreviewFormatter } from "#src/tool-preview-formatter";
import type { PermissionCheckResult } from "#src/types";
// ── helpers ────────────────────────────────────────────────────────────────
function makeFormatter(): ToolPreviewFormatter {
return new ToolPreviewFormatter({
toolInputPreviewMaxLength: TOOL_INPUT_PREVIEW_MAX_LENGTH,
toolTextSummaryMaxLength: TOOL_TEXT_SUMMARY_MAX_LENGTH,
});
}
function makeTcc(overrides: Partial<ToolCallContext> = {}): ToolCallContext {
return {
toolName: "read",
agentName: null,
input: {},
toolCallId: "tc-1",
cwd: "/test/project",
...overrides,
};
}
function makeCheckResult(
state: "allow" | "deny" | "ask",
overrides: Partial<PermissionCheckResult> = {},
): PermissionCheckResult {
return {
state,
toolName: "read",
source: "tool",
origin: "builtin",
matchedPattern: "*",
...overrides,
};
}
// The per-tool gate now receives the AccessPath the pipeline builds, bound to
// the makeTcc default cwd; approval values derive from `accessPath.value()`.
const normalizer = new PathNormalizer(posixPathFlavor, "/test/project");
// ── tests ──────────────────────────────────────────────────────────────────
describe("describeToolGate", () => {
it("gates a standard tool on its own surface and name", () => {
const desc = describeToolGate(
makeTcc({ toolName: "write" }),
makeCheckResult("ask"),
makeFormatter(),
);
expect(desc.surface).toBe("write");
expect(desc.decision).toEqual({ surface: "write", value: "write" });
});
it("returns bash surface with command in decision.value for bash tools", () => {
const check = makeCheckResult("ask", {
toolName: "bash",
command: "git status",
});
const desc = describeToolGate(
makeTcc({ toolName: "bash", input: { command: "git status" } }),
check,
makeFormatter(),
);
expect(desc.surface).toBe("bash");
expect(desc.decision.surface).toBe("bash");
expect(desc.decision.value).toBe("git status");
expect(desc.payload.request.value).toBe("git status");
});
it("gates an aliased shell tool on the bash surface while keeping its tool name in logs (#574)", () => {
const shell: ShellInvocation = {
command: "npm install",
workdir: undefined,
};
const check = makeCheckResult("ask", {
toolName: "bash",
source: "bash",
command: "npm install",
});
const desc = describeToolGate(
makeTcc({ toolName: "exec_command", input: { cmd: "npm install" } }),
check,
makeFormatter(),
undefined,
shell,
);
// Gated as bash: decision, surface, and session rule are bash-shaped.
expect(desc.surface).toBe("bash");
expect(desc.decision.surface).toBe("bash");
expect(desc.decision.value).toBe("npm install");
expect(desc.sessionApproval?.surface).toBe("bash");
expect(desc.sessionApproval?.representativePattern).toBe("npm install*");
// The invoked tool name is preserved for the review log and prompt.
expect(desc.logContext.toolName).toBe("exec_command");
expect(desc.promptDetails.toolName).toBe("exec_command");
// "Gated as bash, invoked as exec_command" is two facts, and the payload
// records both rather than collapsing them.
expect(desc.payload.kind).toBe("bash");
expect(desc.payload.request.toolName).toBe("bash");
expect(desc.payload.request.invokedToolName).toBe("exec_command");
});
it("records a shell alias formatter preview for review and authorization context", () => {
const shell: ShellInvocation = { command: "rm -rf dist", workdir: undefined };
const formatter = new ToolPreviewFormatter(
{
toolInputPreviewMaxLength: TOOL_INPUT_PREVIEW_MAX_LENGTH,
toolTextSummaryMaxLength: TOOL_TEXT_SUMMARY_MAX_LENGTH,
},
{
get: (name) =>
name === "ssh_bash"
? () => "SSH target 'packaging-server' in remote cwd '/srv/build'"
: undefined,
},
);
const desc = describeToolGate(
makeTcc({ toolName: "ssh_bash", input: { command: "rm -rf dist" } }),
makeCheckResult("ask", { toolName: "bash", source: "bash", command: "rm -rf dist" }),
formatter,
undefined,
shell,
);
expect(desc.promptDetails.toolInputPreview).toBe(
"SSH target 'packaging-server' in remote cwd '/srv/build'",
);
expect(desc.logContext.toolInputPreview).toBe(
"SSH target 'packaging-server' in remote cwd '/srv/build'",
);
});
it("returns mcp surface with target in decision.value for MCP tools", () => {
const check = makeCheckResult("ask", {
toolName: "mcp",
target: "server:tool",
});
const desc = describeToolGate(
makeTcc({ toolName: "mcp", input: { tool: "server:tool" } }),
check,
makeFormatter(),
);
expect(desc.surface).toBe("mcp");
expect(desc.decision.surface).toBe("mcp");
expect(desc.decision.value).toBe("server:tool");
});
it("carries the checked tool and its matched rule on the payload", () => {
const check = makeCheckResult("deny", {
toolName: "read",
matchedPattern: "re*",
});
const desc = describeToolGate(makeTcc(), check, makeFormatter());
expect(desc.payload.kind).toBe("tool");
expect(desc.payload.request.toolName).toBe("read");
expect(desc.payload.request.matchedPattern).toBe("re*");
expect(desc.payload.request.requester.agentName).toBeNull();
});
it("binds a current-directory file's session approval to the cwd subtree", () => {
const check = makeCheckResult("ask", { toolName: "edit" });
const desc = describeToolGate(
makeTcc({
toolName: "edit",
input: { path: "index.html" },
cwd: "/test/project",
}),
check,
makeFormatter(),
normalizer.forPath("index.html"),
);
expect(desc.sessionApproval?.surface).toBe("edit");
expect(desc.sessionApproval?.representativePattern).toBe("/test/project/*");
});
it("resolves a sub-directory file's session approval to an absolute pattern", () => {
// The approval value derives from the AccessPath's lexical absolute form
// (`value()`), so sub-directory approvals are absolute too — the deliberate
// tradeoff that keeps the pattern aligned with the policy values it is
// matched against.
const check = makeCheckResult("ask", { toolName: "edit" });
const desc = describeToolGate(
makeTcc({
toolName: "edit",
input: { path: "src/foo.ts" },
cwd: "/test/project",
}),
check,
makeFormatter(),
normalizer.forPath("src/foo.ts"),
);
expect(desc.sessionApproval?.representativePattern).toBe(
"/test/project/src/*",
);
});
it("falls back to a wildcard session approval when no AccessPath is given", () => {
// A path-bearing tool with no `input.path` keeps the `tool` intent and gets
// no AccessPath, so the suggestion collapses to the catch-all.
const desc = describeToolGate(
makeTcc({ toolName: "read", input: {} }),
makeCheckResult("ask"),
makeFormatter(),
);
expect(desc.sessionApproval?.surface).toBe("read");
expect(desc.sessionApproval?.representativePattern).toBe("*");
});
it("populates promptDetails with correct fields", () => {
const check = makeCheckResult("ask");
const desc = describeToolGate(
makeTcc({ toolName: "read", agentName: "my-agent", toolCallId: "tc-42" }),
check,
makeFormatter(),
);
expect(desc.promptDetails).toMatchObject({
source: "tool_call",
agentName: "my-agent",
toolCallId: "tc-42",
toolName: "read",
});
});
it("carries the AccessPath's facts on promptDetails for a path-bearing tool", () => {
const check = makeCheckResult("ask", { toolName: "edit" });
const accessPath = normalizer.forPath("src/foo.ts");
const desc = describeToolGate(
makeTcc({
toolName: "edit",
input: { path: "src/foo.ts" },
cwd: "/test/project",
}),
check,
makeFormatter(),
accessPath,
);
expect(desc.promptDetails.accessIntent).toEqual({
surface: "edit",
matchValues: accessPath.matchValues(),
boundaryValue: accessPath.boundaryValue(),
});
});
it("carries the single decision value on promptDetails for a non-path tool (bash)", () => {
const check = makeCheckResult("ask", {
toolName: "bash",
command: "git status",
});
const desc = describeToolGate(
makeTcc({ toolName: "bash", input: { command: "git status" } }),
check,
makeFormatter(),
);
expect(desc.promptDetails.accessIntent).toEqual({
surface: "bash",
matchValues: ["git status"],
boundaryValue: null,
});
});
it("populates logContext with tool input preview fields", () => {
const check = makeCheckResult("ask", { toolName: "bash", command: "ls" });
const desc = describeToolGate(
makeTcc({ toolName: "bash", input: { command: "ls" } }),
check,
makeFormatter(),
);
expect(desc.logContext).toMatchObject({
source: "tool_call",
toolName: "bash",
});
expect(desc.logContext.command).toBe("ls");
});
});