All Products
Search
Document Center

Cloud Monitor:Digital Employee Skill Format Specification

Last Updated:Mar 01, 2026

This topic describes the format specification for the SKILL.md file and the skill directory structure.

Skill Directory Structure

A complete skill is contained within a folder. The folder name serves as the skill identifier. The directory structure is as follows:

├── SKILL.md        # Required: core instruction file
├── references/     # Optional: reference documentation directory
├── assets/         # Optional: templates and resources directory
└── scripts/        # Optional: executable scripts directory

Directory Descriptions

Directory

Required

Description

Typical Use Cases

SKILL.md

Yes

Core definition file for the skill. Contains metadata and execution instructions.

Defines the skill name, description, execution steps, and output format.

references/

No

Stores reference documents and specifications.

Coding standards, best practices, checklists, API reference, and so on.

assets/

No

Stores template files and static resources.

Output templates, code templates, configuration templates, sample files, and so on.

scripts/

No

Stores executable scripts (not supported yet).

references Directory

The references/ directory stores documentation that the skill requires during execution. The Digital Employee loads these documents on demand as specified in the SKILL.md file.

Use cases:

  • Release checklists and change management guidelines.

  • Runbooks for incident response.

  • Alert handling procedures and escalation paths.

  • CI/CD pipeline configuration guides.

  • Service dependencies and SLO definitions.

Example structure:

references/
├── release-checklist.md     # Release checklist
├── rollback-runbook.md      # Rollback runbook
└── alert-handling.md        # Alert handling guidelines

Reference in SKILL.md:

## Execution Steps
1. Use `references/release-checklist.md` to perform pre-release checks.
2. Use `references/rollback-runbook.md` to assess whether a rollback is needed.

assets Directory

The assets/ directory stores template files and static resources. These resources can be used as references for output formats or as base templates for generating content.

Use cases:

  • Postmortem report templates.

  • Release notification and change record templates.

  • CI/CD pipeline configuration templates.

  • Monitoring and alerting rule templates.

Example structure:

assets/
├── incident-report.md       # Postmortem report template
├── deploy-summary.md        # Release notification template
└── pipeline-config.yaml     # CI/CD pipeline configuration template

Reference in SKILL.md:

## Output Format
Output the postmortem report in the format defined in `assets/incident-report.md`.

## Release Notification
Use `assets/deploy-summary.md` as the content structure for release completion notifications.

scripts Directory

The scripts/ directory stores executable scripts. The Digital Employee can invoke these scripts to automate tasks such as code inspection, data processing, and file transformation.

Note: Script execution is not currently supported in Digital Employee Skills. This directory is reserved for future use.

Use cases:

  • Service health check scripts.

  • Log collection and analysis scripts.

  • CI/CD pipeline trigger and status query scripts.

  • Infrastructure inspection scripts.

Example structure:

scripts/
├── health-check.sh          # Service health check script
├── log-analyzer.py          # Log analysis script
└── pipeline-trigger.sh      # CI/CD pipeline trigger script

Script writing guidelines:

  • Scripts must accept command-line arguments.

  • Output must be structured text for easy parsing by the Digital Employee.

  • Include error handling and usage instructions.

  • Add a shebang on the first line (such as #!/bin/bash or #!/usr/bin/env python3).

SKILL.md File Structure

The SKILL.md file consists of two parts:

  1. YAML Frontmatter: Metadata configuration at the top of the file.

  2. Markdown Body: Specific instructions and descriptions for the skill.

Metadata Fields

Field

Required

Description

Constraints

name

Yes

Unique skill identifier

Lowercase letters, digits, and hyphens only. Maximum 64 characters.

description

Yes

Function description and trigger scenarios

Maximum 512 characters.

Complete Example

---
name: release-readiness-checker
description: Performs readiness checks before service releases. Evaluates health status, traffic, and dependency service readiness.
---

# Release Readiness Checker

## Overview
Before a service release, automatically check key metrics to confirm they meet release criteria. Output a release readiness report to reduce release risk.

## Execution Steps
1. Query current error rate, latency, and QPS baselines for the target service.
2. Check health status of dependent services (database, cache, downstream APIs).
3. Confirm the latest CI/CD pipeline build passed (unit tests and integration testing).
4. Assess current traffic levels to determine if peak business hours apply.
5. Output a release readiness report with flagged risk items.

## Output Format
Output a structured release readiness report that includes:
- Overall readiness conclusion (✅ Ready to release / ⚠️ Proceed with caution / ❌ Not recommended)
- Status and details for each check item
- Risk item explanations and recommended actions

## Notes
- If a metric cannot be retrieved, mark it as "data missing" and skip it. Other checks continue.
- Releases during peak hours (weekdays 10:00–12:00 and 14:00–18:00) require additional confirmation.

Best practices

  • Be precise in descriptions: The description field helps the Digital Employee decide whether to activate the skill. Include clear trigger keywords.

  • Keep steps clear: Use numbered lists to describe execution steps. This makes them easier for the Digital Employee to understand and follow.

  • Provide examples: Include specific input and output examples to clarify expected behavior.

  • Define output formats: Specify the required output format to ensure consistent results.