Merge pull request #3914 from code-yeongyu/fix/max-output-tokens

fix: prevent non-positive maxOutputTokens from leaking to SDK
This commit is contained in:
YeonGyu-Kim
2026-05-11 09:22:39 +09:00
committed by GitHub
4 changed files with 107 additions and 8 deletions
+74 -4
View File
@@ -5,7 +5,7 @@ import { join } from "node:path"
import { createChatParamsHandler, type ChatParamsOutput } from "./chat-params"
import * as dataPathModule from "../shared/data-path"
import { writeProviderModelsCache } from "../shared"
import * as sharedModule from "../shared"
import {
clearSessionPromptParams,
getSessionPromptParams,
@@ -21,13 +21,13 @@ describe("createChatParamsHandler", () => {
getCacheDirSpy = spyOn(dataPathModule, "getOmoOpenCodeCacheDir").mockReturnValue(
join(tempCacheRoot, "oh-my-opencode"),
)
writeProviderModelsCache({ connected: [], models: {} })
sharedModule.writeProviderModelsCache({ connected: [], models: {} })
})
afterEach(() => {
clearSessionPromptParams("ses_chat_params")
clearSessionPromptParams("ses_chat_params_temperature")
writeProviderModelsCache({ connected: [], models: {} })
sharedModule.writeProviderModelsCache({ connected: [], models: {} })
getCacheDirSpy?.mockRestore()
if (tempCacheRoot) {
rmSync(tempCacheRoot, { recursive: true, force: true })
@@ -101,7 +101,7 @@ describe("createChatParamsHandler", () => {
test("applies stored prompt params for the session", async () => {
//#given
writeProviderModelsCache({
sharedModule.writeProviderModelsCache({
connected: ["openai"],
models: {
openai: [
@@ -253,4 +253,74 @@ describe("createChatParamsHandler", () => {
options: {},
})
})
test("falls back to default maxOutputTokens when stored and compatibility tokens are non-positive", async () => {
//#given
const logSpy = spyOn(sharedModule, "log").mockImplementation(() => undefined)
setSessionPromptParams("ses_chat_params", {
maxOutputTokens: 0,
})
const handler = createChatParamsHandler({
anthropicEffort: null,
})
const input = {
sessionID: "ses_chat_params",
agent: { name: "oracle" },
model: { providerID: "custom-provider", modelID: "custom-model" },
provider: { id: "custom-provider" },
message: {},
}
const output: ChatParamsOutput = {
topP: 1,
topK: 1,
maxOutputTokens: 0,
options: {},
}
//#when
await handler(input, output)
//#then
expect(output.maxOutputTokens).toBe(4096)
expect(logSpy).toHaveBeenCalledWith(
"[plugin] maxOutputTokens=0 is non-positive; using safe fallback 4096",
)
logSpy.mockRestore()
})
test("uses safe fallback instead of model max when stored maxOutputTokens is non-positive", async () => {
//#given
setSessionPromptParams("ses_chat_params", {
maxOutputTokens: -1,
})
const handler = createChatParamsHandler({
anthropicEffort: null,
})
const input = {
sessionID: "ses_chat_params",
agent: { name: "oracle" },
model: { providerID: "openai", modelID: "gpt-5.4" },
provider: { id: "openai" },
message: {},
}
const output: ChatParamsOutput = {
topP: 1,
topK: 1,
maxOutputTokens: -1,
options: {},
}
//#when
await handler(input, output)
//#then
expect(output.maxOutputTokens).toBe(4096)
})
})
+17 -4
View File
@@ -1,5 +1,7 @@
import { getSessionPromptParams } from "../shared/session-prompt-params-state"
import { getModelCapabilities, resolveCompatibleModelSettings } from "../shared"
import { getModelCapabilities, log, resolveCompatibleModelSettings } from "../shared"
const SAFE_MAX_OUTPUT_TOKENS_FALLBACK = 4096
export type ChatParamsInput = {
sessionID: string
@@ -96,7 +98,10 @@ export function createChatParamsHandler(args: {
if (storedPromptParams.topP !== undefined) {
output.topP = storedPromptParams.topP
}
if (storedPromptParams.maxOutputTokens !== undefined) {
if (
typeof storedPromptParams.maxOutputTokens === "number" &&
storedPromptParams.maxOutputTokens > 0
) {
(output as Record<string, unknown>).maxOutputTokens = storedPromptParams.maxOutputTokens
}
if (storedPromptParams.options) {
@@ -162,10 +167,18 @@ export function createChatParamsHandler(args: {
}
if ("maxTokens" in compatibility) {
if (compatibility.maxTokens !== undefined) {
if (compatibility.maxTokens !== undefined && compatibility.maxTokens > 0) {
output.maxOutputTokens = compatibility.maxTokens
} else {
delete output.maxOutputTokens
const originalMaxOutputTokens = typeof output.maxOutputTokens === "number"
? output.maxOutputTokens
: compatibility.maxTokens
output.maxOutputTokens = SAFE_MAX_OUTPUT_TOKENS_FALLBACK
if (typeof originalMaxOutputTokens === "number" && originalMaxOutputTokens <= 0) {
log(
`[plugin] maxOutputTokens=${originalMaxOutputTokens} is non-positive; using safe fallback ${SAFE_MAX_OUTPUT_TOKENS_FALLBACK}`,
)
}
}
}
@@ -615,6 +615,18 @@ describe("resolveCompatibleModelSettings", () => {
expect(result.changes).toEqual([])
})
test("#given desired.maxTokens is 0 #then maxTokens is dropped", () => {
const result = resolveCompatibleModelSettings({
providerID: "openai",
modelID: "gpt-5.4",
desired: { maxTokens: 0 },
capabilities: { maxOutputTokens: 128_000 },
})
expect(result.maxTokens).toBeUndefined()
expect(result.changes).toEqual([])
})
// Passthrough: undefined desired values produce no changes
test("no-op when desired settings are empty", () => {
const result = resolveCompatibleModelSettings({
@@ -175,6 +175,10 @@ export function resolveCompatibleModelSettings(
}
let maxTokens = input.desired.maxTokens
if (maxTokens !== undefined && maxTokens <= 0) {
maxTokens = undefined
}
if (
maxTokens !== undefined &&
input.capabilities?.maxOutputTokens !== undefined &&