Writing Skills
writing-skills
效率与协作技能,来自 obra/superpowers,查看原文获取完整用途和操作说明。
byted-contextsearch-operator【部署与运维】Manages Volcano Engine ContextSearch control-plane and application-console operations. Use for ContextSearch scene, AgenticSearch, data source, indexer, search-config, model, deployment, network, specification, and API-key administration. Do not use for generic web search or runtime RAG, image, or video retrieval and chat.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Manage ContextSearch application-console and control-plane operations on Volcano Engine.
Use the bundled CLIs (always run via the skill venv):
{baseDir}/venv/bin/python {baseDir}/scripts/control.py <command>
If {baseDir}/venv does not exist:
python3 -m venv {baseDir}/venv
{baseDir}/venv/bin/pip install -r {baseDir}/requirements.txt
See references/control_plane.md for workflows and examples.
Low-level fallback (use only when goal-based commands do not cover the task):
{baseDir}/venv/bin/python {baseDir}/scripts/contextsearch_cli.py <namespace> <command>
See references/control_tools.md and references/control_coverage.md.
Use goal-based workflows and console action commands to create AgenticSearch and manage scenes, data sources, indexers, search configs, Agentic tools/skills/variables/chats/users, models, deployments, PSR resources, network, specs, and API keys. → See references/control_plane.md.
Low-level tools: Use scripts/contextsearch_cli.py console or openapi only when control.py does not cover a specific console action.
→ See references/control_tools.md.
Common
Execution environment: Always use {baseDir}/venv/bin/python to run scripts.
Control-plane authentication: Console operations require VOLCENGINE_AK and VOLCENGINE_SK. Use VOLCENGINE_REGION only when the default cn-beijing is not correct.
Script usage: Use scripts/control.py for console operations. Use scripts/contextsearch_cli.py only as a last resort when control.py does not cover the console action.
Console parity: For AgenticSearch creation, use control.py create-agentic-search; it follows the console chain ListAgenticSceneTemplate -> CheckBuiltinDeployment -> DeployBuiltinDeployment -> CreateAgenticScene.
Output handling: Commands return JSON. Make decisions from status, goal, data, and steps_completed.
Language (strict): Always reply in the user's language. If the user's message contains Chinese characters, reply in Chinese. Keep commands/flags/code in English.
Console boundary: Treat ContextSearch console operations as control-plane operations, including Agentic data source/indexer/search-config setup.
Destructive actions (strict): Never run delete operations until:
--confirm argument when the goal-based command supports it.Missing parameters: Never fail silently. Inspect available resources or ask one concise question.
Mutation retry safety: If a mutation may have been submitted but the result is unclear, observe with a get/list command before retrying.