# OpenSearch Agent Skills
<a name="opensearch-agent-skills"></a>

Developers today can go from idea to working prototype in minutes using agentic IDEs like Kiro, Claude Code, and Cursor. But whether you're experimenting with a new idea, building a proof of concept, or running production systems, the experience quickly becomes more complex. Search results don't behave as expected, latency spikes require digging through logs, and deploying to AWS introduces configuration decisions that require deep expertise. Translating high-level intent into query DSLs, index configurations, and multi-step workflows still takes significant time – even with an AI agent at your side.

[OpenSearch Agent Skills](https://github.com/opensearch-project/opensearch-agent-skills) addresses this by providing a repository of skills that bring OpenSearch intelligence directly into your agent. Each skill encapsulates domain knowledge, best practices, and multi-step execution logic for a specific workflow – so your agent not only gets results, but understands how they were achieved. Skills work with any coding agent that supports the [Agent Skills standard](https://agentskills.io/specification), including Kiro, Claude Code, and Cursor.

## What you can do with Agent Skills
<a name="agent-skills-use-cases"></a>

### Build a local search application
<a name="agent-skills-use-case-search"></a>

The `opensearch-launchpad` skill brings an intent-driven experience to building and iterating on search applications. It translates natural language requirements or sample data into a fully configured OpenSearch index – with optimized mappings, ingest pipelines, and ML model integrations for keyword, semantic, and hybrid search – and produces a working search experience ready to test and iterate on.

Example prompt:

```
Build a semantic search application for product documentation
```

The agent configures the index mappings, sets up an embedding model and ingest pipeline, and returns a working search API and UI – walking you through each decision along the way.

### Investigate incidents with log analytics
<a name="agent-skills-use-case-logs"></a>

The `log-analytics` skill brings agentic intelligence to observability workflows. Instead of manually crafting Piped Processing Language (PPL) queries or piecing together log data across services, you express your intent and the skill handles the complexity – from error pattern detection to anomaly analysis to root cause correlation.

Example prompt:

```
Investigate why my service is returning 500s and correlate with recent traces
```

The agent runs PPL queries to surface error patterns and log volume anomalies, then correlates log errors to specific trace IDs for faster root cause analysis – without requiring you to master PPL syntax or manually navigate trace data.

### Investigate distributed traces
<a name="agent-skills-use-case-traces"></a>

The `trace-analytics` skill investigates distributed traces directly within OpenSearch. It identifies slow spans, error spans, and service dependencies, and correlates logs and traces using trace IDs to surface root causes across the full observability stack.

Example prompt:

```
Which service is causing the p99 latency regression in the checkout flow?
```

The agent builds a service map, identifies the slow spans, and pinpoints the bottleneck – without requiring you to manually navigate trace data across services.

### Deploy OpenSearch to AWS
<a name="agent-skills-use-case-aws"></a>

The `aws-setup` skill walks your agent through provisioning an OpenSearch Service domain or OpenSearch Serverless collection, configuring access and encryption policies, and connecting your application – with separate guides for managed domains and serverless collections.

Example prompt:

```
Set up an OpenSearch Serverless collection for my search application
```

The agent handles provisioning, policy configuration, and validation, guiding you through each step.

## Available skills
<a name="agent-skills-available"></a>

Skills are organized by category. You can install the complete collection or individual skills:


| Category | Skill | What it does | 
| --- | --- | --- | 
| Search | opensearch-launchpad | Scaffolds a search application from scratch. Covers BM25 lexical search, semantic search, hybrid search, and agentic search. Includes model selection guides, evaluation strategies, and sample data. | 
| Observability | log-analytics | Queries and analyzes logs using Piped Processing Language (PPL). Helps the agent detect error patterns, surface anomalies, and correlate events across indexes. | 
| Observability | trace-analytics | Investigates distributed traces. Identifies slow spans, builds service maps, and traces AI agent tool invocations through your system. | 
| Cloud | aws-setup | Deploys OpenSearch to AWS. Includes dedicated guides for Amazon OpenSearch Service domains and OpenSearch Serverless collections, covering provisioning, access policies, and network configuration. | 

The project is under active development. For the latest skill list and contribution guidance, see the [opensearch-agent-skills](https://github.com/opensearch-project/opensearch-agent-skills) repository.

## Prerequisites
<a name="agent-skills-prerequisites"></a>

Prerequisites vary by skill. To use the full collection you need:
+ A coding agent that supports Agent Skills (Claude Code, Cursor, or Kiro).
+ [Node.js](https://nodejs.org/) with `npx` on your `PATH`, to run the `skills` installer.
+ [https://docs.astral.sh/uv/getting-started/installation/](https://docs.astral.sh/uv/getting-started/installation/) and Python 3.11 or later. Skills use `uv` to run scripts in isolated environments.
+ [Docker](https://docs.docker.com/get-docker/) installed and running, if you plan to use skills that spin up a local cluster (such as `opensearch-launchpad`).
+ AWS credentials configured with appropriate permissions if you plan to use the `aws-setup` skill. For guidance on required permissions, see [Identity-based policies](ac.md#ac-types-identity).

## Install and use skills
<a name="agent-skills-install"></a>

Skills install through the [https://agentskills.io](https://agentskills.io) CLI, which you invoke with `npx`. The installer detects your agent and places skill files in the correct location.

To install the entire collection:

```
npx skills add opensearch-project/opensearch-agent-skills
```

To install a single skill, use the `@skill-name` suffix. The `--full-depth` flag includes all reference material the agent may need for follow-up questions:

```
npx skills add opensearch-project/opensearch-agent-skills@opensearch-launchpad --full-depth
npx skills add opensearch-project/opensearch-agent-skills@log-analytics --full-depth
npx skills add opensearch-project/opensearch-agent-skills@trace-analytics --full-depth
npx skills add opensearch-project/opensearch-agent-skills@aws-setup --full-depth
```

Common install options:

**`-a {{agent-name}}`**  
Install to a specific agent, for example `-a claude-code`.

**`-g`**  
Install globally so the skill is available across all projects on your machine.

**`--all`**  
Install to every agent the CLI detects on your machine.

After installation, restart your agent so it picks up the new skill files. Then try a prompt such as *"I want to build a hybrid search app with OpenSearch."* The agent reads the skill instructions and runs the required scripts directly.

The `aws-setup` skill includes dedicated guides for both OpenSearch Service product families. For Amazon OpenSearch Service managed domains, the skill walks through creating a domain, configuring access policies, and connecting your application (see [Introduction](what-is.md)). For OpenSearch Serverless, it covers creating a collection, configuring data access and encryption policies, and ingesting data (see [Amazon OpenSearch Serverless](serverless.md)). For interactive queries against a running cluster, pair the skills with the [OpenSearch MCP server](opensearch-mcp-server.md).

## Security considerations
<a name="agent-skills-security"></a>

When you install and run a skill, your agent executes the scripts and commands it describes. Treat skills as you would any third-party developer tool:
+ Review the skill's `SKILL.md` file in the [source repository](https://github.com/opensearch-project/opensearch-agent-skills) before installing.
+ Run skills in a development or sandbox environment before pointing them at production resources.
+ Use least-privilege IAM credentials when running the `aws-setup` skill, scoped to the AWS account and AWS Region where you intend to deploy.
+ Review any AWS resources the skill creates before committing to long-lived infrastructure. Skills may provision domains, collections, IAM roles, and networking resources on your behalf.

## Additional resources
<a name="agent-skills-more-information"></a>
+ [opensearch-agent-skills](https://github.com/opensearch-project/opensearch-agent-skills) on GitHub – Source, issues, and the developer guide for authoring new skills.
+ [Agent Skills specification](https://agentskills.io/specification) – The open standard that skills implement.
+ [OpenSearch MCP server](opensearch-mcp-server.md) – The companion MCP server for interactive queries against a running cluster.