Harness Skills

Claude Code skills for the Harness.io CI/CD platform. Generate pipeline YAML, manage resources, debug failures, analyze costs, and more from natural language.

This repository is designed as a workflow system, not just a folder of prompts. The top-level instructions (CLAUDE.md, AGENTS.md, .github/copilot-instructions.md) establish shared behavior, while individual skills specialize in creation, debugging, governance, and reporting tasks.

Prerequisites

Harness MCP v2 Server - required for MCP-powered skills. Most skills in this repo depend on it for Harness API access.

Setup

Claude Code

Clone the repo and run Claude Code from the project directory. Skills are automatically discovered from CLAUDE.md and skills/*/SKILL.md:

hljs language-bash

git clone https://github.com/harness/harness-skills.git
cd harness-skills
claude

To add the Harness MCP server, configure it in your Claude Code settings (~/.claude/settings.json):

hljs language-json

{
  "mcpServers": {
    "harness-mcp-v2": {
      "command": "npx",
      "args": ["-y", "harness-mcp-v2"],
      "env": {
        "HARNESS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Invoke skills by name:

hljs language-css

/create-pipeline
Create a CI pipeline for a Node.js app that builds, runs tests,
and pushes a Docker image to ECR

Cursor

Clone the repo into your project or as a reference workspace:

hljs language-bash

git clone https://github.com/harness/harness-skills.git

The repo includes .cursor/rules/harness.mdc, which Cursor automatically loads as a project rule.
Configure the Harness MCP server in Cursor (~/.cursor/mcp.json):

hljs language-json

{
  "mcpServers": {
    "harness-mcp-v2": {
      "command": "npx",
      "args": ["-y", "harness-mcp-v2"],
      "env": {
        "HARNESS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Reference individual skills in your prompts using @file:

hljs language-sql

@harness-skills/skills/create-pipeline/SKILL.md
Create a CI pipeline for my Go service

OpenAI Codex

Clone the repo into your working directory:

hljs language-bash

git clone https://github.com/harness/harness-skills.git

The repo includes AGENTS.md at the root, which Codex automatically reads as system instructions.
Configure the Harness MCP server in your Codex MCP config:

hljs language-json

{
  "mcpServers": {
    "harness-mcp-v2": {
      "command": "npx",
      "args": ["-y", "harness-mcp-v2"],
      "env": {
        "HARNESS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Reference individual skill files as context when prompting:

hljs language-swift

Using the instructions in harness-skills/skills/debug-pipeline/SKILL.md,
diagnose why my deploy pipeline failed

GitHub Copilot

Clone the repo into your project:

hljs language-bash

git clone https://github.com/harness/harness-skills.git

The repo includes .github/copilot-instructions.md, which Copilot automatically reads as project-level context in both GitHub.com and VS Code.
For VS Code, configure the Harness MCP server in your workspace settings (.vscode/mcp.json):

hljs language-json

{
  "servers": {
    "harness-mcp-v2": {
      "command": "npx",
      "args": ["-y", "harness-mcp-v2"],
      "env": {
        "HARNESS_API_KEY": "<your-api-key>"
      }
    }
  }
}

Reference skill files in Copilot Chat using #file:

hljs language-swift

#file:harness-skills/skills/create-pipeline/SKILL.md
Create a CI pipeline for my Python service

For GitHub Copilot on GitHub.com, attach skill files as context in Copilot Chat or add them as knowledge base references in your Copilot organization settings.

Windsurf / Other AI Editors

The skills in this repo are plain Markdown files with YAML frontmatter. They work with any AI coding tool that supports:

System instructions - Use CLAUDE.md as project-level context.
MCP servers - Connect the Harness MCP v2 server for API access.
File context - Reference individual skills/*/SKILL.md files in prompts.

Operating Model

The best Harness skills follow the same control flow even when they target different resource types:

Establish scope first - confirm account/org/project context before listing, creating, updating, or deleting resources.
Verify dependencies before generating dependents - do not reference connectors, secrets, environments, infrastructure, or templates that have not been confirmed to exist.
Discover schema before writing payloads - use harness_describe and API validation feedback instead of guessing field names or payload shape.

These repo-level playbooks live in:

references/scope-establishment.md
references/dependency-check-playbook.md
references/schema-validation-loop.md
templates/operation-summary.md

Workflow Modes

Workflow mode	Representative skills	Use when
Create and scaffold	`/create-pipeline`, `/create-service`, `/create-connector`, `/create-template`	You need to define or generate new Harness resources and their YAML or MCP payloads.
Run and debug	`/run-pipeline`, `/debug-pipeline`, `/migrate-pipeline`, `/manage-delegates`	You already have resources and need to execute, diagnose, or repair behavior.
Govern and secure	`/manage-roles`, `/manage-users`, `/create-policy`, `/security-report`, `/audit-report`	You need RBAC, policy, compliance, or security workflows with blast-radius awareness.
Analyze and report	`/dora-metrics`, `/analyze-costs`, `/scorecard-review`, `/template-usage`	You need structured reports, summaries, recommendations, or adoption analysis.

End-to-End Workflows

New microservice setup

Use these skills in order:

/create-connector
/create-secret
/create-service
/create-environment
/create-infrastructure
/create-pipeline
/create-trigger

Debug a failed deployment

Typical sequence:

/run-pipeline to identify the latest execution or reproduce the issue
/debug-pipeline to classify the failure and inspect root cause
/template-usage if shared templates may have propagated the issue
/manage-delegates if the failure points to delegate capacity or connectivity

Skills

Pipeline & Template Creation

Skill	Description
`/create-pipeline`	Generate v0 Pipeline YAML (CI, CD, approvals, matrix strategies)
`/create-pipeline-v1`	Generate v1 simplified Pipeline YAML - Alpha: internal testing only
`/create-template`	Create reusable Step, Stage, Pipeline, or StepGroup templates
`/create-trigger`	Create webhook, scheduled, and artifact triggers
`/create-agent-template`	Create AI-powered agent templates - Alpha: internal testing only

Resource Management

Skill	Description
`/create-service`	Create service definitions (K8s, Helm, ECS, Lambda)
`/create-environment`	Create environment definitions with overrides
`/create-infrastructure`	Create infrastructure definitions
`/create-connector`	Create connectors (Git, cloud, registries, clusters)
`/create-secret`	Create secrets (text, file, SSH, WinRM)

Access Control & Feature Flags (MCP)

Skill	Description
`/manage-users`	Manage users, user groups, and service accounts
`/manage-roles`	Manage role assignments and RBAC
`/manage-feature-flags`	Create, list, toggle, and delete feature flags

Operations & Debugging (MCP)

Skill	Description
`/run-pipeline`	Execute pipelines, monitor progress, handle approvals
`/debug-pipeline`	Analyze execution failures, diagnose root causes
`/migrate-pipeline`	Convert pipelines from v0 to v1 format
`/optimize-pipeline`	Pipeline speed optimization, parallel testing, caching, monorepo builds
`/deployment-readiness`	Pre-deployment readiness checks, environment drift, canary decisions
`/incident-response`	Deployment-incident correlation, blast radius, postmortems
`/pr-analysis`	PR pipeline impact, security review, PR-to-production tracking
`/template-usage`	Track template dependencies and adoption
`/manage-delegates`	Monitor delegate health and manage tokens

Security

Skill	Description
`/configure-repo-scan`	Scan your code repositories for SAST and SCA vulnerabilities
`/configure-secret-scan`	Detect exposed secrets and credentials in your code repositories
`/configure-container-scan`	Scan your container images for vulnerabilities
`/configure-dast-scan`	Test your running applications for security vulnerabilities with DAST
`/exempt-vuln`	Exempt vulnerabilities with security waivers at Target, project or pipeline scope
`/approve-exempt`	Approve pending STO exemptions at current scope or elevate to project/org/account; mixed scopes per batch
`/generate-slsa`	Generate and attest SLSA provenance for your artifacts
`/enforce-slsa`	Verify SLSA provenance attestations and enforce OPA policies on your artifacts
`/create-sbom`	Generate and attest SBOM for your artifacts and code repositories
`/enforce-sbom`	Verify SBOM attestations and enforce OPA policies on your artifacts and code repositories
`/sign-artifact`	Sign your artifacts to ensure integrity and authenticity in your supply chain
`/verify-sign`	Verify artifact integrity and authenticity before deployment

Platform Intelligence (MCP)

Skill	Description
`/analyze-costs`	Cloud cost analysis and optimization (CCM)
`/security-report`	Vulnerability reports, SBOMs, compliance (SCS/STO)
`/dora-metrics`	DORA metrics and engineering performance (SEI)
`/sei-analytics`	Sprint analytics, investment allocation, capacity forecasting, release readiness
`/manage-slos`	SLO definition, error budgets, incident detection, runbooks (SRM)
`/ai-operations`	Predictive failure analysis and intelligent alert correlation (AIDA)
`/gitops-status`	GitOps application health and sync status
`/chaos-experiment`	Create and run chaos experiments
`/scorecard-review`	Service maturity scorecards (IDP)
`/manage-idp`	Service catalog, self-service workflows, documentation, onboarding (IDP)
`/manage-iacm`	Terraform workspaces, drift detection, cost estimation (IaCM)
`/manage-cde`	Cloud development environments and workspace templates (CDE)
`/manage-artifacts`	Artifact registry, security scanning, replication (AR)
`/manage-supply-chain`	SBOM generation, artifact signing, supply chain policies (SSCA)
`/audit-report`	Audit trails and compliance reports
`/create-policy`	Create OPA governance policies for supply chain security

Project Structure

hljs language-bash

harness-skills/
├── skills/
│   ├── create-pipeline/
│   │   ├── SKILL.md
│   │   └── references/
│   ├── create-template/
│   │   └── SKILL.md
│   ├── debug-pipeline/
│   │   └── SKILL.md
│   └── ...                      # Skill definitions
├── references/                  # Shared repo-level playbooks
├── templates/                   # Shared repo-level output templates
├── scripts/
│   └── validate-skills.sh       # Structural validation
├── examples/
│   ├── v0/                      # v0 pipeline examples
│   ├── v1/                      # v1 pipeline examples
│   ├── templates/               # Template examples
│   ├── triggers/                # Trigger examples
│   ├── services/                # Service definition examples
│   ├── environments/            # Environment examples
│   ├── connectors/              # Connector examples
│   └── ...
├── .cursor/rules/harness.mdc    # Auto-loaded by Cursor
├── .github/copilot-instructions.md
├── AGENTS.md                    # Auto-loaded by OpenAI Codex
├── CLAUDE.md                    # Auto-loaded by Claude Code
├── CONTRIBUTING.md              # Contribution guidelines
├── LICENSE
└── README.md

Shared Playbooks and Templates

Use root-level shared assets for behavior that should stay consistent across many skills:

references/ - reusable operating guidance such as scope establishment, dependency verification, and schema-validation loops
templates/ - reusable output contracts such as operation summaries and report formats

Use per-skill references/ or templates/ when the content is domain-specific and should not be imported broadly.

Skill Anatomy

Each skill is a directory under skills/ containing a SKILL.md with YAML frontmatter and a consistent markdown body:

hljs language-yaml

---
name: my-skill
description: >-
  Explain what the skill does, when to use it, when not to use it, and likely
  trigger phrases. Keep it under 1024 characters.
metadata:
  author: Harness
  version: 1.0.0
  mcp-server: harness-mcp-v2
license: Apache-2.0
compatibility: Requires Harness MCP v2 server (harness-mcp-v2)
---

# My Skill

One or two sentences describing the operating mode and expected outcome.

## Instructions

Phase-based steps for Claude to follow.

## Examples

Invocation examples and, for complex skills, brief worked examples.

## Performance Notes

Validation checks, tradeoffs, or speed/accuracy guidance.

## Troubleshooting

Common errors, recovery steps, and expected fallbacks.

For complex skills, add references/ or templates/ under the skill directory rather than bloating SKILL.md.

MCP Tools

MCP-powered skills use the Harness MCP v2 server, which provides 10 generic tools dispatched by resource_type:

Tool	Purpose
`harness_list`	List resources
`harness_get`	Get resource details
`harness_create`	Create a resource
`harness_update`	Update a resource
`harness_delete`	Delete a resource
`harness_execute`	Execute an action
`harness_search`	Search across resources
`harness_describe`	Get resource schema
`harness_diagnose`	Diagnose issues
`harness_status`	Check system status

Schema References

Contributing

See CONTRIBUTING.md for skill authoring standards, validation rules, and contribution workflow.

License

This project is licensed under the Apache License 2.0.

harness-skills

Harness Skills

Prerequisites

Setup

Claude Code

Cursor

OpenAI Codex

GitHub Copilot

Windsurf / Other AI Editors

Operating Model

Workflow Modes

End-to-End Workflows

New microservice setup

Debug a failed deployment

Skills

Pipeline & Template Creation

Resource Management

Access Control & Feature Flags (MCP)

Operations & Debugging (MCP)

Security

Platform Intelligence (MCP)

Project Structure

Shared Playbooks and Templates

Skill Anatomy

MCP Tools

Schema References

Contributing

License

Similar Packages

Tags

Original Author

Publisher