mirror of
https://github.com/openclaw/openclaw.git
synced 2026-05-06 10:20:42 +00:00
fix(hooks): harden cli transcript loading (#70786)
This commit is contained in:
@@ -18,6 +18,7 @@ import {
|
||||
} from "./cli-runner.test-support.js";
|
||||
import { executePreparedCliRun } from "./cli-runner/execute.js";
|
||||
import { resolveCliNoOutputTimeoutMs } from "./cli-runner/helpers.js";
|
||||
import * as sessionHistoryModule from "./cli-runner/session-history.js";
|
||||
import { MAX_CLI_SESSION_HISTORY_MESSAGES } from "./cli-runner/session-history.js";
|
||||
import type { PreparedCliRunContext } from "./cli-runner/types.js";
|
||||
|
||||
@@ -510,6 +511,34 @@ describe("runCliAgent reliability", () => {
|
||||
}
|
||||
});
|
||||
|
||||
it("does not emit llm_output when the CLI run returns no assistant text", async () => {
|
||||
const hookRunner = {
|
||||
hasHooks: vi.fn((hookName: string) => hookName === "llm_output"),
|
||||
runLlmInput: vi.fn(async () => undefined),
|
||||
runLlmOutput: vi.fn(async () => undefined),
|
||||
runAgentEnd: vi.fn(async () => undefined),
|
||||
};
|
||||
mockGetGlobalHookRunner.mockReturnValue(hookRunner as never);
|
||||
|
||||
supervisorSpawnMock.mockResolvedValueOnce(
|
||||
createManagedRun({
|
||||
reason: "exit",
|
||||
exitCode: 0,
|
||||
exitSignal: null,
|
||||
durationMs: 50,
|
||||
stdout: " ",
|
||||
stderr: "",
|
||||
timedOut: false,
|
||||
noOutputTimedOut: false,
|
||||
}),
|
||||
);
|
||||
|
||||
const result = await runPreparedCliAgent(buildPreparedContext());
|
||||
|
||||
expect(result.payloads).toBeUndefined();
|
||||
expect(hookRunner.runLlmOutput).not.toHaveBeenCalled();
|
||||
});
|
||||
|
||||
it("emits agent_end with failure details when the CLI run fails", async () => {
|
||||
const hookRunner = {
|
||||
hasHooks: vi.fn((hookName: string) => ["llm_input", "agent_end"].includes(hookName)),
|
||||
@@ -634,6 +663,41 @@ describe("runCliAgent reliability", () => {
|
||||
fs.rmSync(dir, { recursive: true, force: true });
|
||||
}
|
||||
});
|
||||
|
||||
it("skips transcript loading when only llm_output hooks are active", async () => {
|
||||
const hookRunner = {
|
||||
hasHooks: vi.fn((hookName: string) => hookName === "llm_output"),
|
||||
runLlmInput: vi.fn(async () => undefined),
|
||||
runLlmOutput: vi.fn(async () => undefined),
|
||||
runAgentEnd: vi.fn(async () => undefined),
|
||||
};
|
||||
mockGetGlobalHookRunner.mockReturnValue(hookRunner as never);
|
||||
const historySpy = vi.spyOn(sessionHistoryModule, "loadCliSessionHistoryMessages");
|
||||
|
||||
supervisorSpawnMock.mockResolvedValueOnce(
|
||||
createManagedRun({
|
||||
reason: "exit",
|
||||
exitCode: 0,
|
||||
exitSignal: null,
|
||||
durationMs: 50,
|
||||
stdout: "hello from cli",
|
||||
stderr: "",
|
||||
timedOut: false,
|
||||
noOutputTimedOut: false,
|
||||
}),
|
||||
);
|
||||
|
||||
try {
|
||||
await runPreparedCliAgent(buildPreparedContext());
|
||||
|
||||
expect(historySpy).not.toHaveBeenCalled();
|
||||
await vi.waitFor(() => {
|
||||
expect(hookRunner.runLlmOutput).toHaveBeenCalledTimes(1);
|
||||
});
|
||||
} finally {
|
||||
historySpy.mockRestore();
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
describe("resolveCliNoOutputTimeoutMs", () => {
|
||||
|
||||
@@ -1,7 +1,9 @@
|
||||
import { formatErrorMessage } from "../infra/errors.js";
|
||||
import { getGlobalHookRunner } from "../plugins/hook-runner-global.js";
|
||||
import { loadCliSessionHistoryMessages } from "./cli-runner/session-history.js";
|
||||
import type { PreparedCliRunContext, RunCliAgentParams } from "./cli-runner/types.js";
|
||||
import { FailoverError, isFailoverError, resolveFailoverStatus } from "./failover-error.js";
|
||||
import { buildAgentHookConversationMessages } from "./harness/hook-history.js";
|
||||
import {
|
||||
runAgentHarnessAgentEndHook,
|
||||
runAgentHarnessLlmInputHook,
|
||||
@@ -53,13 +55,20 @@ export async function runPreparedCliAgent(
|
||||
): Promise<EmbeddedPiRunResult> {
|
||||
const { executePreparedCliRun } = await import("./cli-runner/execute.runtime.js");
|
||||
const { params } = context;
|
||||
const historyMessages = loadCliSessionHistoryMessages({
|
||||
sessionId: params.sessionId,
|
||||
sessionFile: params.sessionFile,
|
||||
sessionKey: params.sessionKey,
|
||||
agentId: params.agentId,
|
||||
config: params.config,
|
||||
});
|
||||
const hookRunner = getGlobalHookRunner();
|
||||
const hasLlmInputHooks = hookRunner?.hasHooks("llm_input") === true;
|
||||
const hasLlmOutputHooks = hookRunner?.hasHooks("llm_output") === true;
|
||||
const hasAgentEndHooks = hookRunner?.hasHooks("agent_end") === true;
|
||||
const historyMessages =
|
||||
hasLlmInputHooks || hasAgentEndHooks
|
||||
? loadCliSessionHistoryMessages({
|
||||
sessionId: params.sessionId,
|
||||
sessionFile: params.sessionFile,
|
||||
sessionKey: params.sessionKey,
|
||||
agentId: params.agentId,
|
||||
config: params.config,
|
||||
})
|
||||
: [];
|
||||
const llmInputEvent = {
|
||||
runId: params.runId,
|
||||
sessionId: params.sessionId,
|
||||
@@ -82,9 +91,13 @@ export async function runPreparedCliAgent(
|
||||
} as const;
|
||||
|
||||
const buildAgentEndMessages = (lastAssistant?: unknown): unknown[] => [
|
||||
...historyMessages,
|
||||
buildCliHookUserMessage(params.prompt),
|
||||
...(lastAssistant ? [lastAssistant] : []),
|
||||
...buildAgentHookConversationMessages({
|
||||
historyMessages,
|
||||
currentTurnMessages: [
|
||||
buildCliHookUserMessage(params.prompt),
|
||||
...(lastAssistant ? [lastAssistant] : []),
|
||||
],
|
||||
}),
|
||||
];
|
||||
|
||||
const buildFailedAgentEndEvent = (error: string) => ({
|
||||
@@ -125,18 +138,20 @@ export async function runPreparedCliAgent(
|
||||
usage: output.usage,
|
||||
})
|
||||
: undefined;
|
||||
runAgentHarnessLlmOutputHook({
|
||||
event: {
|
||||
runId: params.runId,
|
||||
sessionId: params.sessionId,
|
||||
provider: params.provider,
|
||||
model: context.modelId,
|
||||
assistantTexts,
|
||||
...(lastAssistant ? { lastAssistant } : {}),
|
||||
...(output.usage ? { usage: output.usage } : {}),
|
||||
},
|
||||
ctx: hookContext,
|
||||
});
|
||||
if (assistantText.length > 0 && hasLlmOutputHooks) {
|
||||
runAgentHarnessLlmOutputHook({
|
||||
event: {
|
||||
runId: params.runId,
|
||||
sessionId: params.sessionId,
|
||||
provider: params.provider,
|
||||
model: context.modelId,
|
||||
assistantTexts,
|
||||
...(lastAssistant ? { lastAssistant } : {}),
|
||||
...(output.usage ? { usage: output.usage } : {}),
|
||||
},
|
||||
ctx: hookContext,
|
||||
});
|
||||
}
|
||||
return { output, assistantText, lastAssistant };
|
||||
};
|
||||
|
||||
|
||||
@@ -123,6 +123,41 @@ describe("loadCliSessionHistoryMessages", () => {
|
||||
}
|
||||
});
|
||||
|
||||
it("rejects symlinked transcripts instead of following them outside the sessions directory", () => {
|
||||
const stateDir = fs.mkdtempSync(path.join(os.tmpdir(), "openclaw-cli-state-"));
|
||||
const outsideDir = fs.mkdtempSync(path.join(os.tmpdir(), "openclaw-cli-outside-"));
|
||||
vi.stubEnv("OPENCLAW_STATE_DIR", stateDir);
|
||||
const canonicalSessionFile = path.join(
|
||||
stateDir,
|
||||
"agents",
|
||||
"main",
|
||||
"sessions",
|
||||
"session-symlink.jsonl",
|
||||
);
|
||||
const outsideFile = createSessionTranscript({
|
||||
rootDir: outsideDir,
|
||||
sessionId: "session-symlink",
|
||||
filePath: path.join(outsideDir, "outside.jsonl"),
|
||||
messages: ["stolen history"],
|
||||
});
|
||||
fs.mkdirSync(path.dirname(canonicalSessionFile), { recursive: true });
|
||||
fs.symlinkSync(outsideFile, canonicalSessionFile);
|
||||
|
||||
try {
|
||||
expect(
|
||||
loadCliSessionHistoryMessages({
|
||||
sessionId: "session-symlink",
|
||||
sessionFile: canonicalSessionFile,
|
||||
sessionKey: "agent:main:main",
|
||||
agentId: "main",
|
||||
}),
|
||||
).toEqual([]);
|
||||
} finally {
|
||||
fs.rmSync(stateDir, { recursive: true, force: true });
|
||||
fs.rmSync(outsideDir, { recursive: true, force: true });
|
||||
}
|
||||
});
|
||||
|
||||
it("drops oversized transcript files instead of loading them into hook payloads", () => {
|
||||
const stateDir = fs.mkdtempSync(path.join(os.tmpdir(), "openclaw-cli-state-"));
|
||||
vi.stubEnv("OPENCLAW_STATE_DIR", stateDir);
|
||||
@@ -149,4 +184,37 @@ describe("loadCliSessionHistoryMessages", () => {
|
||||
fs.rmSync(stateDir, { recursive: true, force: true });
|
||||
}
|
||||
});
|
||||
|
||||
it("honors custom session store roots when resolving hook history transcripts", () => {
|
||||
const stateDir = fs.mkdtempSync(path.join(os.tmpdir(), "openclaw-cli-state-"));
|
||||
const customStoreDir = fs.mkdtempSync(path.join(os.tmpdir(), "openclaw-cli-store-"));
|
||||
vi.stubEnv("OPENCLAW_STATE_DIR", stateDir);
|
||||
const storePath = path.join(customStoreDir, "sessions.json");
|
||||
fs.writeFileSync(storePath, "{}", "utf-8");
|
||||
const sessionFile = createSessionTranscript({
|
||||
rootDir: customStoreDir,
|
||||
sessionId: "session-custom-store",
|
||||
filePath: path.join(customStoreDir, "session-custom-store.jsonl"),
|
||||
messages: ["custom store history"],
|
||||
});
|
||||
|
||||
try {
|
||||
expect(
|
||||
loadCliSessionHistoryMessages({
|
||||
sessionId: "session-custom-store",
|
||||
sessionFile,
|
||||
sessionKey: "agent:main:main",
|
||||
agentId: "main",
|
||||
config: {
|
||||
session: {
|
||||
store: storePath,
|
||||
},
|
||||
},
|
||||
}),
|
||||
).toMatchObject([{ role: "user", content: "custom store history" }]);
|
||||
} finally {
|
||||
fs.rmSync(stateDir, { recursive: true, force: true });
|
||||
fs.rmSync(customStoreDir, { recursive: true, force: true });
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
@@ -1,4 +1,5 @@
|
||||
import fs from "node:fs";
|
||||
import path from "node:path";
|
||||
import { SessionManager } from "@mariozechner/pi-coding-agent";
|
||||
import {
|
||||
resolveSessionFilePath,
|
||||
@@ -6,9 +7,26 @@ import {
|
||||
} from "../../config/sessions/paths.js";
|
||||
import type { OpenClawConfig } from "../../config/types.openclaw.js";
|
||||
import { resolveSessionAgentIds } from "../agent-scope.js";
|
||||
import {
|
||||
limitAgentHookHistoryMessages,
|
||||
MAX_AGENT_HOOK_HISTORY_MESSAGES,
|
||||
} from "../harness/hook-history.js";
|
||||
|
||||
export const MAX_CLI_SESSION_HISTORY_FILE_BYTES = 5 * 1024 * 1024;
|
||||
export const MAX_CLI_SESSION_HISTORY_MESSAGES = 200;
|
||||
export const MAX_CLI_SESSION_HISTORY_MESSAGES = MAX_AGENT_HOOK_HISTORY_MESSAGES;
|
||||
|
||||
function safeRealpathSync(filePath: string): string | undefined {
|
||||
try {
|
||||
return fs.realpathSync(filePath);
|
||||
} catch {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
function isPathWithinBase(basePath: string, targetPath: string): boolean {
|
||||
const relative = path.relative(basePath, targetPath);
|
||||
return Boolean(relative) && !relative.startsWith("..") && !path.isAbsolute(relative);
|
||||
}
|
||||
|
||||
function resolveSafeCliSessionFile(params: {
|
||||
sessionId: string;
|
||||
@@ -16,19 +34,25 @@ function resolveSafeCliSessionFile(params: {
|
||||
sessionKey?: string;
|
||||
agentId?: string;
|
||||
config?: OpenClawConfig;
|
||||
}): string {
|
||||
}): { sessionFile: string; sessionsDir: string } {
|
||||
const { defaultAgentId, sessionAgentId } = resolveSessionAgentIds({
|
||||
sessionKey: params.sessionKey,
|
||||
config: params.config,
|
||||
agentId: params.agentId,
|
||||
});
|
||||
return resolveSessionFilePath(
|
||||
const pathOptions = resolveSessionFilePathOptions({
|
||||
agentId: sessionAgentId ?? defaultAgentId,
|
||||
storePath: params.config?.session?.store,
|
||||
});
|
||||
const sessionFile = resolveSessionFilePath(
|
||||
params.sessionId,
|
||||
{ sessionFile: params.sessionFile },
|
||||
resolveSessionFilePathOptions({
|
||||
agentId: sessionAgentId ?? defaultAgentId,
|
||||
}),
|
||||
pathOptions,
|
||||
);
|
||||
return {
|
||||
sessionFile,
|
||||
sessionsDir: pathOptions?.sessionsDir ?? path.dirname(sessionFile),
|
||||
};
|
||||
}
|
||||
|
||||
export function loadCliSessionHistoryMessages(params: {
|
||||
@@ -39,28 +63,25 @@ export function loadCliSessionHistoryMessages(params: {
|
||||
config?: OpenClawConfig;
|
||||
}): unknown[] {
|
||||
try {
|
||||
const sessionFile = resolveSafeCliSessionFile(params);
|
||||
if (!fs.existsSync(sessionFile)) {
|
||||
const { sessionFile, sessionsDir } = resolveSafeCliSessionFile(params);
|
||||
const entryStat = fs.lstatSync(sessionFile);
|
||||
if (!entryStat.isFile() || entryStat.isSymbolicLink()) {
|
||||
return [];
|
||||
}
|
||||
const stat = fs.statSync(sessionFile);
|
||||
const realSessionsDir = safeRealpathSync(sessionsDir) ?? path.resolve(sessionsDir);
|
||||
const realSessionFile = safeRealpathSync(sessionFile);
|
||||
if (!realSessionFile || !isPathWithinBase(realSessionsDir, realSessionFile)) {
|
||||
return [];
|
||||
}
|
||||
const stat = fs.statSync(realSessionFile);
|
||||
if (!stat.isFile() || stat.size > MAX_CLI_SESSION_HISTORY_FILE_BYTES) {
|
||||
return [];
|
||||
}
|
||||
const entries = SessionManager.open(sessionFile).getEntries();
|
||||
const history: unknown[] = [];
|
||||
for (let index = entries.length - 1; index >= 0; index -= 1) {
|
||||
const entry = entries[index];
|
||||
if (entry?.type !== "message") {
|
||||
continue;
|
||||
}
|
||||
history.push(entry.message as unknown);
|
||||
if (history.length >= MAX_CLI_SESSION_HISTORY_MESSAGES) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
history.reverse();
|
||||
return history;
|
||||
const entries = SessionManager.open(realSessionFile).getEntries();
|
||||
const history = entries.flatMap((entry) =>
|
||||
entry?.type === "message" ? [entry.message as unknown] : [],
|
||||
);
|
||||
return limitAgentHookHistoryMessages(history, MAX_CLI_SESSION_HISTORY_MESSAGES);
|
||||
} catch {
|
||||
return [];
|
||||
}
|
||||
|
||||
@@ -3,6 +3,11 @@ export type PluginEntryConfig = {
|
||||
hooks?: {
|
||||
/** Controls prompt mutation via before_prompt_build and prompt fields from legacy before_agent_start. */
|
||||
allowPromptInjection?: boolean;
|
||||
/**
|
||||
* Controls access to raw conversation content from llm_input/llm_output/agent_end hooks.
|
||||
* Non-bundled plugins must opt in explicitly; bundled plugins stay allowed unless disabled.
|
||||
*/
|
||||
allowConversationAccess?: boolean;
|
||||
};
|
||||
subagent?: {
|
||||
/** Explicitly allow this plugin to request per-run provider/model overrides for subagent runs. */
|
||||
|
||||
@@ -21,6 +21,7 @@ export type NormalizedPluginsConfig = {
|
||||
enabled?: boolean;
|
||||
hooks?: {
|
||||
allowPromptInjection?: boolean;
|
||||
allowConversationAccess?: boolean;
|
||||
};
|
||||
subagent?: {
|
||||
allowModelOverride?: boolean;
|
||||
@@ -80,12 +81,21 @@ function normalizePluginEntries(
|
||||
? {
|
||||
allowPromptInjection: (hooksRaw as { allowPromptInjection?: unknown })
|
||||
.allowPromptInjection,
|
||||
allowConversationAccess: (hooksRaw as { allowConversationAccess?: unknown })
|
||||
.allowConversationAccess,
|
||||
}
|
||||
: undefined;
|
||||
const normalizedHooks =
|
||||
hooks && typeof hooks.allowPromptInjection === "boolean"
|
||||
hooks &&
|
||||
(typeof hooks.allowPromptInjection === "boolean" ||
|
||||
typeof hooks.allowConversationAccess === "boolean")
|
||||
? {
|
||||
allowPromptInjection: hooks.allowPromptInjection,
|
||||
...(typeof hooks.allowPromptInjection === "boolean"
|
||||
? { allowPromptInjection: hooks.allowPromptInjection }
|
||||
: {}),
|
||||
...(typeof hooks.allowConversationAccess === "boolean"
|
||||
? { allowConversationAccess: hooks.allowConversationAccess }
|
||||
: {}),
|
||||
}
|
||||
: undefined;
|
||||
const subagentRaw = entry.subagent;
|
||||
|
||||
@@ -70,10 +70,12 @@ describe("normalizePluginsConfig", () => {
|
||||
entry: {
|
||||
hooks: {
|
||||
allowPromptInjection: false,
|
||||
allowConversationAccess: true,
|
||||
},
|
||||
},
|
||||
expectedHooks: {
|
||||
allowPromptInjection: false,
|
||||
allowConversationAccess: true,
|
||||
},
|
||||
},
|
||||
{
|
||||
@@ -81,7 +83,8 @@ describe("normalizePluginsConfig", () => {
|
||||
entry: {
|
||||
hooks: {
|
||||
allowPromptInjection: "nope",
|
||||
} as unknown as { allowPromptInjection: boolean },
|
||||
allowConversationAccess: "nope",
|
||||
} as unknown as { allowPromptInjection: boolean; allowConversationAccess: boolean },
|
||||
},
|
||||
expectedHooks: undefined,
|
||||
},
|
||||
|
||||
@@ -138,6 +138,19 @@ const promptInjectionHookNameSet = new Set<PluginHookName>(PROMPT_INJECTION_HOOK
|
||||
export const isPromptInjectionHookName = (hookName: PluginHookName): boolean =>
|
||||
promptInjectionHookNameSet.has(hookName);
|
||||
|
||||
export const CONVERSATION_HOOK_NAMES = [
|
||||
"llm_input",
|
||||
"llm_output",
|
||||
"agent_end",
|
||||
] as const satisfies readonly PluginHookName[];
|
||||
|
||||
export type ConversationHookName = (typeof CONVERSATION_HOOK_NAMES)[number];
|
||||
|
||||
const conversationHookNameSet = new Set<PluginHookName>(CONVERSATION_HOOK_NAMES);
|
||||
|
||||
export const isConversationHookName = (hookName: PluginHookName): boolean =>
|
||||
conversationHookNameSet.has(hookName);
|
||||
|
||||
export type PluginHookAgentContext = {
|
||||
runId?: string;
|
||||
agentId?: string;
|
||||
|
||||
@@ -5303,6 +5303,67 @@ module.exports = {
|
||||
]);
|
||||
});
|
||||
|
||||
it("blocks conversation typed hooks for non-bundled plugins unless explicitly allowed", () => {
|
||||
useNoBundledPlugins();
|
||||
const plugin = writePlugin({
|
||||
id: "conversation-hooks",
|
||||
filename: "conversation-hooks.cjs",
|
||||
body: `module.exports = { id: "conversation-hooks", register(api) {
|
||||
api.on("llm_input", () => undefined);
|
||||
api.on("llm_output", () => undefined);
|
||||
api.on("agent_end", () => undefined);
|
||||
} };`,
|
||||
});
|
||||
|
||||
const registry = loadRegistryFromSinglePlugin({
|
||||
plugin,
|
||||
pluginConfig: {
|
||||
allow: ["conversation-hooks"],
|
||||
},
|
||||
});
|
||||
|
||||
expect(registry.typedHooks).toEqual([]);
|
||||
const blockedDiagnostics = registry.diagnostics.filter((diag) =>
|
||||
diag.message.includes(
|
||||
"non-bundled plugins must set plugins.entries.conversation-hooks.hooks.allowConversationAccess=true",
|
||||
),
|
||||
);
|
||||
expect(blockedDiagnostics).toHaveLength(3);
|
||||
});
|
||||
|
||||
it("allows conversation typed hooks for non-bundled plugins when explicitly enabled", () => {
|
||||
useNoBundledPlugins();
|
||||
const plugin = writePlugin({
|
||||
id: "conversation-hooks-allowed",
|
||||
filename: "conversation-hooks-allowed.cjs",
|
||||
body: `module.exports = { id: "conversation-hooks-allowed", register(api) {
|
||||
api.on("llm_input", () => undefined);
|
||||
api.on("llm_output", () => undefined);
|
||||
api.on("agent_end", () => undefined);
|
||||
} };`,
|
||||
});
|
||||
|
||||
const registry = loadRegistryFromSinglePlugin({
|
||||
plugin,
|
||||
pluginConfig: {
|
||||
allow: ["conversation-hooks-allowed"],
|
||||
entries: {
|
||||
"conversation-hooks-allowed": {
|
||||
hooks: {
|
||||
allowConversationAccess: true,
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
});
|
||||
|
||||
expect(registry.typedHooks.map((entry) => entry.hookName)).toEqual([
|
||||
"llm_input",
|
||||
"llm_output",
|
||||
"agent_end",
|
||||
]);
|
||||
});
|
||||
|
||||
it("ignores unknown typed hooks from plugins and keeps loading", () => {
|
||||
useNoBundledPlugins();
|
||||
const plugin = writePlugin({
|
||||
|
||||
@@ -83,6 +83,7 @@ import { withPluginRuntimePluginIdScope } from "./runtime/gateway-request-scope.
|
||||
import type { PluginRuntime } from "./runtime/types.js";
|
||||
import { defaultSlotIdForKey, hasKind } from "./slots.js";
|
||||
import {
|
||||
isConversationHookName,
|
||||
isPluginHookName,
|
||||
isPromptInjectionHookName,
|
||||
stripPromptMutationFieldsFromLegacyHookResult,
|
||||
@@ -165,6 +166,7 @@ export type {
|
||||
|
||||
type PluginTypedHookPolicy = {
|
||||
allowPromptInjection?: boolean;
|
||||
allowConversationAccess?: boolean;
|
||||
};
|
||||
|
||||
const constrainLegacyPromptInjectionHook = (
|
||||
@@ -1210,6 +1212,29 @@ export function createPluginRegistry(registryParams: PluginRegistryParams) {
|
||||
) as PluginHookHandlerMap[K];
|
||||
}
|
||||
}
|
||||
if (isConversationHookName(hookName)) {
|
||||
const explicitConversationAccess = policy?.allowConversationAccess;
|
||||
if (record.origin !== "bundled" && explicitConversationAccess !== true) {
|
||||
pushDiagnostic({
|
||||
level: "warn",
|
||||
pluginId: record.id,
|
||||
source: record.source,
|
||||
message:
|
||||
`typed hook "${hookName}" blocked because non-bundled plugins must set ` +
|
||||
`plugins.entries.${record.id}.hooks.allowConversationAccess=true`,
|
||||
});
|
||||
return;
|
||||
}
|
||||
if (record.origin === "bundled" && explicitConversationAccess === false) {
|
||||
pushDiagnostic({
|
||||
level: "warn",
|
||||
pluginId: record.id,
|
||||
source: record.source,
|
||||
message: `typed hook "${hookName}" blocked by plugins.entries.${record.id}.hooks.allowConversationAccess=false`,
|
||||
});
|
||||
return;
|
||||
}
|
||||
}
|
||||
record.hookCount += 1;
|
||||
registry.typedHooks.push({
|
||||
pluginId: record.id,
|
||||
|
||||
@@ -466,6 +466,7 @@ describe("plugin status reports", () => {
|
||||
expect(inspect).not.toBeNull();
|
||||
expectInspectPolicy(inspect!, {
|
||||
allowPromptInjection: undefined,
|
||||
allowConversationAccess: undefined,
|
||||
allowModelOverride: true,
|
||||
allowedModels: ["openai/gpt-5.5"],
|
||||
hasAllowedModelsConfig: true,
|
||||
@@ -583,7 +584,7 @@ describe("plugin status reports", () => {
|
||||
plugins: {
|
||||
entries: {
|
||||
google: {
|
||||
hooks: { allowPromptInjection: false },
|
||||
hooks: { allowPromptInjection: false, allowConversationAccess: true },
|
||||
subagent: {
|
||||
allowModelOverride: true,
|
||||
allowedModels: ["openai/gpt-5.5"],
|
||||
@@ -623,6 +624,7 @@ describe("plugin status reports", () => {
|
||||
]);
|
||||
expectInspectPolicy(inspect!, {
|
||||
allowPromptInjection: false,
|
||||
allowConversationAccess: true,
|
||||
allowModelOverride: true,
|
||||
allowedModels: ["openai/gpt-5.5"],
|
||||
hasAllowedModelsConfig: true,
|
||||
|
||||
@@ -82,6 +82,7 @@ export type PluginInspectReport = {
|
||||
diagnostics: PluginDiagnostic[];
|
||||
policy: {
|
||||
allowPromptInjection?: boolean;
|
||||
allowConversationAccess?: boolean;
|
||||
allowModelOverride?: boolean;
|
||||
allowedModels: string[];
|
||||
hasAllowedModelsConfig: boolean;
|
||||
@@ -348,6 +349,7 @@ export function buildPluginInspectReport(params: {
|
||||
diagnostics,
|
||||
policy: {
|
||||
allowPromptInjection: policyEntry?.hooks?.allowPromptInjection,
|
||||
allowConversationAccess: policyEntry?.hooks?.allowConversationAccess,
|
||||
allowModelOverride: policyEntry?.subagent?.allowModelOverride,
|
||||
allowedModels: [...(policyEntry?.subagent?.allowedModels ?? [])],
|
||||
hasAllowedModelsConfig: policyEntry?.subagent?.hasAllowedModelsConfig === true,
|
||||
|
||||
Reference in New Issue
Block a user