一键导入
create-ops-agent
Create a standardized operational agent for a component — orchestrates the full analyze → design → generate → validate workflow
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Create a standardized operational agent for a component — orchestrates the full analyze → design → generate → validate workflow
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
Generate a deepagents-based operational agent project from templates and design document
Run AST guard validation on a generated agent project and auto-fix violations
Analyze a component's codebase to understand its logging, metrics, and infrastructure for ops agent creation
Design an operational agent based on codebase analysis, determining tools, config, and system prompt content
基于 SOC 职业分类
| name | create-ops-agent |
| description | Create a standardized operational agent for a component — orchestrates the full analyze → design → generate → validate workflow |
You are helping a component owner create a standardized operational agent using deepagents. This skill orchestrates the entire process.
You will walk the owner through 4 phases. Each phase has its own detailed skill — read and follow them in order. Do NOT skip phases or combine them.
Phase 1: Analyze → Understand the component's codebase
Phase 2: Design → Decide what the agent needs
Phase 3: Generate → Create the agent project
Phase 4: Validate → Verify structure and run tests
skills/ — the 4 phase skillstemplates/ — agent project boilerplatetools/ — reference tool implementations (OpenSearch, Prometheus)guard/ — AST structure validatorRead and follow skills/01-analyze-codebase/SKILL.md.
Goal: Understand the component's logging, metrics, infrastructure dependencies, and failure patterns.
Output: <component>-analysis.md
Gate: Confirm with the owner that the analysis is accurate before proceeding.
"Analysis complete. Ready to move on to designing your agent?"
Read and follow skills/02-design-agent/SKILL.md.
Goal: Determine tool list, config values, and system prompt content.
Output: <component>-design.md
Gate: Owner approves the design.
"Design complete. Ready to generate the agent project?"
Read and follow skills/03-generate-agent/SKILL.md.
Goal: Create a complete, working agent project from templates.
Output: <component>-ops-agent/ directory with all files
Gate: Guard validation passes.
"Agent project generated. Running structure validation..."
Read and follow skills/04-validate-structure/SKILL.md.
Goal: Verify the generated project follows all structural rules and tests pass.
Output: Clean guard report + passing tests
Gate: All checks pass.
After all 4 phases pass, present the final summary:
"Your operational agent is ready!
Project:
<component>-ops-agent/Quick start:
cd <component>-ops-agent pip install -e '.[dev]' python agent.pyWhat it can do:
- [tool list with descriptions]
To add tools later:
- Create a new file in
tools/(one public function + docstring)- Add config section in
config/agent.yaml- Export from
tools/__init__.pyand register inagent.py- Run
python -m guard check .to verifyTo re-validate:
python -m guard check <component>-ops-agent/"
python -m guard check.