fix(delegate-task): default run_in_background and load_skills instead of throwing (fixes #4119)

Sisyphus and other delegators occasionally invoke the task() tool without
an explicit run_in_background or load_skills argument. The runtime
validators in tool-argument-preparation.ts threw a hard Error in that
case, which short-circuited tool.execute() entirely. Because OpenCode's
tool.execute.after hook only runs on returned results, the
delegate-task-retry hook never had a chance to attach corrective
guidance — so the model saw a raw failure and either burned several
retries or fell back to a synchronous Explore call, silently losing
parallel execution.

Behavior change:
- run_in_background omitted -> defaults to false (sync delegation), with
  a log entry for observability.
- load_skills omitted or null -> normalized to [] with a log entry on
  the explicit-null path.
- The Zod schema entries are now .optional() and their .describe()
  strings declare the defaults honestly; the markdown tool description
  was updated to match (no more 'REQUIRED' lie).

The orthogonal validation 'Must provide either category or
subagent_type.' is unchanged and still surfaces as a returned error.

Tests:
- The five throw-on-missing tests in tools.test.ts are rewritten to
  assert the new default-and-proceed contract.
- The 'no category, no subagent_type' test now asserts the
  missing-target error remains intact.

Refs the workaround the reporter validated in the original issue body;
matches the design from PR #2375 which was previously reverted by
566031f4.
This commit is contained in:
Claude Agent
2026-05-17 19:47:17 +02:00
parent 49066e7ecc
commit 2f16a7da9f
4 changed files with 159 additions and 122 deletions
@@ -29,9 +29,17 @@ export async function prepareDelegateTaskArgs(args: Record<string, unknown>, ctx
title: description, title: description,
}) })
const runInBackground = args.run_in_background let runInBackground = args.run_in_background
if (runInBackground === undefined) { if (runInBackground === undefined) {
throw new Error("Invalid arguments: 'run_in_background' parameter is REQUIRED. Specify run_in_background=false for task delegation, or run_in_background=true for parallel exploration.") // Default to sync delegation. Tool description still nudges the model to be
// explicit, but a missing flag should not fail an otherwise valid call —
// hard-failing here burns turns and silently downgrades parallel work to
// synchronous fallbacks. See issue #4119.
runInBackground = false
log("[task] run_in_background omitted; defaulting to false (sync delegation)", {
category: args.category,
subagent_type: originalSubagentType,
})
} }
let loadSkills = args.load_skills let loadSkills = args.load_skills
@@ -44,12 +52,16 @@ export async function prepareDelegateTaskArgs(args: Record<string, unknown>, ctx
} }
} }
if (loadSkills === undefined) { if (loadSkills === undefined || loadSkills === null) {
throw new Error("Invalid arguments: 'load_skills' parameter is REQUIRED. Pass [] if no skills needed.") // Default to no skills. Same rationale as run_in_background above: callers
} // that omit the field already implicitly mean "no skill content needed".
if (loadSkills === null) {
if (loadSkills === null) { log("[task] load_skills=null received; normalizing to []", {
throw new Error("Invalid arguments: load_skills=null is not allowed. Pass [] if no skills needed.") category: args.category,
subagent_type: originalSubagentType,
})
}
loadSkills = []
} }
const normalizedLoadSkills = Array.isArray(loadSkills) const normalizedLoadSkills = Array.isArray(loadSkills)
+6 -6
View File
@@ -42,17 +42,17 @@ export function createDelegateTaskPresentation(options: DelegateTaskToolOptions)
**COMMON MISTAKE (DO NOT DO THIS):** **COMMON MISTAKE (DO NOT DO THIS):**
\`\`\` \`\`\`
task(description="...", prompt="...", run_in_background=false) // ❌ FAILS - missing category AND subagent_type task(description="...", prompt="...") // ❌ FAILS - missing category AND subagent_type
\`\`\` \`\`\`
**CORRECT - Using category:** **CORRECT - Using category:**
\`\`\` \`\`\`
task(category="quick", load_skills=[], description="Fix type error", prompt="...", run_in_background=false) task(category="quick", description="Fix type error", prompt="...")
\`\`\` \`\`\`
**CORRECT - Using subagent_type:** **CORRECT - Using subagent_type with parallel exploration:**
\`\`\` \`\`\`
task(subagent_type="explore", load_skills=[], description="Find patterns", prompt="...", run_in_background=true) task(subagent_type="explore", description="Find patterns", prompt="...", run_in_background=true)
\`\`\` \`\`\`
REQUIRED: Provide ONE of: REQUIRED: Provide ONE of:
@@ -61,12 +61,12 @@ export function createDelegateTaskPresentation(options: DelegateTaskToolOptions)
**DO NOT provide both.** If category is provided, subagent_type is ignored. **DO NOT provide both.** If category is provided, subagent_type is ignored.
- load_skills: ALWAYS REQUIRED. Pass [] if no skills needed, or ["skill-1", "skill-2"] for category tasks. - load_skills: Optional. Defaults to [] when omitted. Pass ["skill-1", "skill-2"] for skill-specific tasks.
- category: Use predefined category → Spawns Sisyphus-Junior with category config - category: Use predefined category → Spawns Sisyphus-Junior with category config
Available categories: Available categories:
${categoryList} ${categoryList}
- subagent_type: Use specific agent directly (explore, librarian, oracle, metis, momus) - subagent_type: Use specific agent directly (explore, librarian, oracle, metis, momus)
- run_in_background: REQUIRED. true=async (returns a background task ID like \`bg_...\` for \`background_output\`), false=sync (waits). Use background=true ONLY for parallel exploration with 5+ independent queries. - run_in_background: Optional. Defaults to false (sync, waits). Set true=async (returns a background task ID like \`bg_...\` for \`background_output\`) ONLY for parallel exploration with 5+ independent queries.
Sync waits use a 30-minute inactivity window: OpenCode busy/retry/running status resets the window, so this is not a total wall-clock limit. Sync waits use a 30-minute inactivity window: OpenCode busy/retry/running status resets the window, so this is not a total wall-clock limit.
- task_id: Continuation session id (\`ses_...\`) from task metadata. Continues the same subagent session with FULL CONTEXT PRESERVED; not the background task id (\`bg_...\`). - task_id: Continuation session id (\`ses_...\`) from task metadata. Continues the same subagent session with FULL CONTEXT PRESERVED; not the background task id (\`bg_...\`).
- command: The command that triggered this task (optional, for slash command tracking). - command: The command that triggered this task (optional, for slash command tracking).
+116 -95
View File
@@ -1180,19 +1180,30 @@ describe("sisyphus-task", () => {
}) })
describe("skills parameter", () => { describe("skills parameter", () => {
test("skills parameter is required - throws error when not provided", async () => { test("#given load_skills omitted #when executing #then defaults to [] and proceeds (fixes #4119)", async () => {
// given // given
const { createDelegateTask } = require("./tools") const { createDelegateTask } = require("./tools")
let promptBody: any
const mockManager = { launch: async () => ({}) } const mockManager = { launch: async () => ({}) }
const promptMock = async (input: any) => {
promptBody = input.body
return { data: {} }
}
const mockClient = { const mockClient = {
app: { agents: async () => ({ data: [] }) }, app: { agents: async () => ({ data: [] }) },
config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) }, config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) },
session: { session: {
get: async () => ({ data: { directory: "/project" } }),
create: async () => ({ data: { id: "test-session" } }), create: async () => ({ data: { id: "test-session" } }),
prompt: async () => ({ data: {} }), prompt: promptMock,
promptAsync: async () => ({ data: {} }), promptAsync: promptMock,
messages: async () => ({ data: [] }), messages: async () => ({
data: [{ info: { role: "assistant" }, parts: [{ type: "text", text: "Done" }] }]
}),
status: async () => ({ data: {} }),
}, },
} }
@@ -1208,9 +1219,9 @@ describe("sisyphus-task", () => {
abort: new AbortController().signal, abort: new AbortController().signal,
} }
// when - skills not provided (undefined) // when - skills not provided (undefined); previously threw a hard error.
// then - should throw error about missing skills // then - should default to [] and proceed normally.
await expect(tool.execute( await tool.execute(
{ {
description: "Test task", description: "Test task",
prompt: "Do something", prompt: "Do something",
@@ -1218,50 +1229,65 @@ describe("sisyphus-task", () => {
run_in_background: false, run_in_background: false,
}, },
toolContext toolContext
)).rejects.toThrow("Invalid arguments: 'load_skills' parameter is REQUIRED") )
})
test("null skills throws error", async () => { expect(promptBody).toBeDefined()
// given }, { timeout: 20000 })
const { createDelegateTask } = require("./tools")
const mockManager = { launch: async () => ({}) } test("#given load_skills=null #when executing #then normalizes to [] and proceeds (fixes #4119)", async () => {
const mockClient = { // given
app: { agents: async () => ({ data: [] }) }, const { createDelegateTask } = require("./tools")
config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) }, let promptBody: any
session: {
create: async () => ({ data: { id: "test-session" } }),
prompt: async () => ({ data: {} }),
promptAsync: async () => ({ data: {} }),
messages: async () => ({ data: [] }),
},
}
const tool = createDelegateTask({ const mockManager = { launch: async () => ({}) }
manager: mockManager,
client: mockClient,
})
const toolContext = { const promptMock = async (input: any) => {
sessionID: "parent-session", promptBody = input.body
messageID: "parent-message", return { data: {} }
agent: "sisyphus", }
abort: new AbortController().signal,
}
// when - null passed const mockClient = {
// then - should throw error about null app: { agents: async () => ({ data: [] }) },
await expect(tool.execute( config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) },
{ session: {
description: "Test task", get: async () => ({ data: { directory: "/project" } }),
prompt: "Do something", create: async () => ({ data: { id: "test-session" } }),
category: "ultrabrain", prompt: promptMock,
run_in_background: false, promptAsync: promptMock,
load_skills: null, messages: async () => ({
}, data: [{ info: { role: "assistant" }, parts: [{ type: "text", text: "Done" }] }]
toolContext }),
)).rejects.toThrow("Invalid arguments: load_skills=null is not allowed") status: async () => ({ data: {} }),
}) },
}
const tool = createDelegateTask({
manager: mockManager,
client: mockClient,
})
const toolContext = {
sessionID: "parent-session",
messageID: "parent-message",
agent: "sisyphus",
abort: new AbortController().signal,
}
// when - load_skills=null passed; previously threw "load_skills=null is not allowed".
// then - should normalize to [] and proceed.
await tool.execute(
{
description: "Test task",
prompt: "Do something",
category: "ultrabrain",
run_in_background: false,
load_skills: null,
},
toolContext
)
expect(promptBody).toBeDefined()
}, { timeout: 20000 })
test("empty array [] is allowed and proceeds without skill content", async () => { test("empty array [] is allowed and proceeds without skill content", async () => {
// given // given
@@ -1320,25 +1346,33 @@ describe("sisyphus-task", () => {
}) })
describe("run_in_background parameter", () => { describe("run_in_background parameter", () => {
test("#given category without run_in_background #when executing #then throws required parameter error", async () => { test("#given category without run_in_background #when executing #then defaults to sync and proceeds (fixes #4119)", async () => {
// given // given
const { createDelegateTask } = require("./tools") const { createDelegateTask } = require("./tools")
let promptBody: any
const promptMock = async (input: any) => {
promptBody = input.body
return { data: {} }
}
const mockManager = { launch: async () => ({}) } const mockManager = { launch: async () => ({}) }
const mockClient = { const mockClient = {
app: { agents: async () => ({ data: [] }) }, app: { agents: async () => ({ data: [] }) },
config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) }, config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) },
session: { session: {
get: async () => ({ data: { directory: "/project" } }),
create: async () => ({ data: { id: "test-session" } }), create: async () => ({ data: { id: "test-session" } }),
prompt: async () => ({ data: {} }), prompt: promptMock,
promptAsync: async () => ({ data: {} }), promptAsync: promptMock,
messages: async () => ({ data: [] }), messages: async () => ({
data: [{ info: { role: "assistant" }, parts: [{ type: "text", text: "Done" }] }]
}),
status: async () => ({ data: {} }),
}, },
} }
const tool = createDelegateTask({ manager: mockManager, client: mockClient }) const tool = createDelegateTask({ manager: mockManager, client: mockClient })
// when // when - run_in_background omitted (previously a hard throw)
// then await tool.execute(
await expect(tool.execute(
{ {
description: "Category without run flag", description: "Category without run flag",
prompt: "Do something", prompt: "Do something",
@@ -1346,28 +1380,39 @@ describe("sisyphus-task", () => {
load_skills: [], load_skills: [],
}, },
{ sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal } { sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal }
)).rejects.toThrow("Invalid arguments: 'run_in_background' parameter is REQUIRED") )
})
test("#given subagent_type without run_in_background #when executing #then throws required parameter error", async () => { // then - sync path should run and the session prompt should be sent
expect(promptBody).toBeDefined()
}, { timeout: 20000 })
test("#given subagent_type without run_in_background #when executing #then defaults to sync and proceeds (fixes #4119)", async () => {
// given // given
const { createDelegateTask } = require("./tools") const { createDelegateTask } = require("./tools")
let promptBody: any
const promptMock = async (input: any) => {
promptBody = input.body
return { data: {} }
}
const mockManager = { launch: async () => ({}) } const mockManager = { launch: async () => ({}) }
const mockClient = { const mockClient = {
app: { agents: async () => ({ data: [{ name: "explore", mode: "subagent" }] }) }, app: { agents: async () => ({ data: [{ name: "explore", mode: "subagent" }] }) },
config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) }, config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) },
session: { session: {
get: async () => ({ data: { directory: "/project" } }),
create: async () => ({ data: { id: "test-session" } }), create: async () => ({ data: { id: "test-session" } }),
prompt: async () => ({ data: {} }), prompt: promptMock,
promptAsync: async () => ({ data: {} }), promptAsync: promptMock,
messages: async () => ({ data: [] }), messages: async () => ({
data: [{ info: { role: "assistant" }, parts: [{ type: "text", text: "Done" }] }]
}),
status: async () => ({ data: {} }),
}, },
} }
const tool = createDelegateTask({ manager: mockManager, client: mockClient }) const tool = createDelegateTask({ manager: mockManager, client: mockClient })
// when // when
// then await tool.execute(
await expect(tool.execute(
{ {
description: "Subagent without run flag", description: "Subagent without run flag",
prompt: "Find patterns", prompt: "Find patterns",
@@ -1375,39 +1420,13 @@ describe("sisyphus-task", () => {
load_skills: [], load_skills: [],
}, },
{ sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal } { sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal }
)).rejects.toThrow("Invalid arguments: 'run_in_background' parameter is REQUIRED") )
})
test("#given task_id without run_in_background #when executing #then throws required parameter error", async () => {
// given
const { createDelegateTask } = require("./tools")
const mockManager = { resume: async () => ({ id: "task-1", sessionId: "ses_1", status: "running" }) }
const mockClient = {
app: { agents: async () => ({ data: [] }) },
config: { get: async () => ({ data: { model: SYSTEM_DEFAULT_MODEL } }) },
session: {
create: async () => ({ data: { id: "test-session" } }),
prompt: async () => ({ data: {} }),
promptAsync: async () => ({ data: {} }),
messages: async () => ({ data: [] }),
},
}
const tool = createDelegateTask({ manager: mockManager, client: mockClient })
// when
// then // then
await expect(tool.execute( expect(promptBody).toBeDefined()
{ }, { timeout: 20000 })
description: "Continue without run flag",
prompt: "Continue",
task_id: "ses_existing",
load_skills: [],
},
{ sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal }
)).rejects.toThrow("Invalid arguments: 'run_in_background' parameter is REQUIRED")
})
test("#given no category no subagent_type no task_id and no run_in_background #when executing #then throws required parameter error", async () => { test("#given no category no subagent_type and no run_in_background #when executing #then still returns the (different) missing-target error (fixes #4119)", async () => {
// given // given
const { createDelegateTask } = require("./tools") const { createDelegateTask } = require("./tools")
const mockManager = { launch: async () => ({}) } const mockManager = { launch: async () => ({}) }
@@ -1423,16 +1442,18 @@ describe("sisyphus-task", () => {
} }
const tool = createDelegateTask({ manager: mockManager, client: mockClient }) const tool = createDelegateTask({ manager: mockManager, client: mockClient })
// when // when - omitting run_in_background no longer throws, but missing category+subagent_type still produces a (returned) error.
// then const result = await tool.execute(
await expect(tool.execute(
{ {
description: "Missing required args", description: "Missing required args",
prompt: "Do something", prompt: "Do something",
load_skills: [], load_skills: [],
}, },
{ sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal } { sessionID: "parent-session", messageID: "parent-message", agent: "sisyphus", abort: new AbortController().signal }
)).rejects.toThrow("Invalid arguments: 'run_in_background' parameter is REQUIRED") )
// then - the missing-target error remains intact; only the run_in_background gate was removed.
expect(String(result)).toContain("Must provide either category or subagent_type")
}) })
test("#given category without description #when executing #then auto-generates description from prompt", async () => { test("#given category without description #when executing #then auto-generates description from prompt", async () => {
+6 -2
View File
@@ -21,12 +21,16 @@ export type { SyncSessionCreatedEvent, DelegateTaskToolOptions, BuildSystemConte
export { buildSystemContent, buildTaskPrompt } from "./prompt-builder" export { buildSystemContent, buildTaskPrompt } from "./prompt-builder"
const delegateTaskArgsSchema = { const delegateTaskArgsSchema = {
load_skills: tool.schema.array(tool.schema.string()).describe("Skill names to inject. REQUIRED - pass [] if no skills needed."), load_skills: tool.schema
.array(tool.schema.string())
.optional()
.describe("Skill names to inject. Optional; defaults to [] when omitted. Pass an explicit array (e.g. [\"git-master\"]) for skill-specific tasks."),
description: tool.schema.string().optional().describe("Short task description (3-5 words). Auto-generated from prompt if omitted."), description: tool.schema.string().optional().describe("Short task description (3-5 words). Auto-generated from prompt if omitted."),
prompt: tool.schema.string().describe("Full detailed prompt for the agent"), prompt: tool.schema.string().describe("Full detailed prompt for the agent"),
run_in_background: tool.schema run_in_background: tool.schema
.boolean() .boolean()
.describe("REQUIRED. true=async (returns background task ID `bg_...` for background_output), false=sync (waits). Use false for task delegation, true ONLY for parallel exploration."), .optional()
.describe("Optional; defaults to false (sync). true=async (returns background task ID `bg_...` for background_output), false=sync (waits). Use true ONLY for parallel exploration; otherwise omit or pass false for task delegation."),
category: tool.schema.string().optional().describe("REQUIRED if subagent_type not provided. Do NOT provide both category and subagent_type."), category: tool.schema.string().optional().describe("REQUIRED if subagent_type not provided. Do NOT provide both category and subagent_type."),
subagent_type: tool.schema.string().optional().describe("REQUIRED if category not provided. Do NOT provide both category and subagent_type."), subagent_type: tool.schema.string().optional().describe("REQUIRED if category not provided. Do NOT provide both category and subagent_type."),
task_id: tool.schema task_id: tool.schema