Bailian Subagent

Spawn a bailian (百炼 DashScope) subagent to handle token-heavy or compute-intensive tasks. Use when the main agent needs to delegate data processing, batch operations, DataWorks/MaxCompute SQL execution, news fetching, or any task where saving Claude tokens matters. Triggers on phrases like "讓百炼做", "丟給百炼", "省token", "subagent去做", "delegate to bailian", or when a task is large/repetitive enough to warrant offloading.

Delegate heavy tasks to a cheaper bailian model to save Claude tokens.

Available Models

Model	Use case
bailian/qwen3.5-plus	General tasks, data processing, writing
bailian/qwen3-coder-plus	Code generation, debugging, scripts
bailian/kimi-k2.5	Long context, document analysis
bailian/glm-5	Alternative general model

Default recommendation: bailian/qwen3.5-plus for most tasks.

How to Spawn

text

sessions_spawn(
    task="<detailed instructions>",
    model="bailian/qwen3.5-plus",
    runtime="subagent",
    mode="run",
    runTimeoutSeconds=600
)

What to Delegate

Good candidates:

• DataWorks / MaxCompute SQL execution (DROP/CREATE/INSERT/SELECT)
• Batch data processing or transformation
• Fetching and summarizing news / RSS feeds
• Writing boilerplate code or scripts
• Long document analysis or summarization
• Any task with large context or repetitive steps

Keep with Claude (main agent):

• Final decisions and trading advice
• Conversation with Samuel
• Tasks requiring memory/context of prior sessions
• Sensitive judgment calls

Writing Good Task Instructions

Include all required context in the task string — subagent starts fresh with no memory:

• Credentials / endpoints (if needed)
• Exact steps to follow
• Expected output format
• Error handling instructions

DataWorks Pattern (verified)

When delegating MC/DW SQL tasks, provide this pattern in the task:

python

from alibabacloud_dataworks_public20240518.client import Client
from alibabacloud_dataworks_public20240518 import models as dw_models
from alibabacloud_tea_openapi.models import Config
import uuid, time

config = Config(
    access_key_id='${ALICLOUD_ACCESS_KEY_ID}',
    access_key_secret='${ALICLOUD_ACCESS_KEY_SECRET}',
    endpoint='dataworks.cn-hangzhou.aliyuncs.com',
    region_id='cn-hangzhou'
)
client = Client(config)
# ... run_sql() helper pattern

Key params: project_id=579810, owner=1206989323863181, resource_group_id=Serverless_res_group_421503158374466_777119112505568, datasource=samuelhsin

Sleep 15s between SQL calls to avoid throttling.