Enrich Port AI with skills
Skills give Port AI domain-specific instructions at runtime. Instead of repeating the same prompts or hoping the model infers your process, you package expert workflows as skills that Port AI loads when a task matches.
This page covers how skills enrich Port AI during a conversation or agent run. For skill concepts, blueprints, and creation steps, see the Skills overview.
When to use skills
Use skills when you want Port AI to follow a repeatable process, for example:
- Troubleshooting integration sync issues with a fixed diagnostic sequence.
- Guiding deployment decisions against your organization's approval rules.
- Running incident root-cause analysis with consistent data-gathering steps.
Skills work across Port AI Assistant, AI agents, the Port MCP Server, and the AI chat widget.
How Port AI loads skills
Port AI exposes a load_skill tool through the Port MCP Server. When your prompt or agent task matches a skill's description, Port AI loads that skill before continuing.
load_skill({ name: "troubleshoot-integration" })
You can also load a specific reference file from a skill:
load_skill({ name: "troubleshoot-integration", resource: "references/common-errors.md" })
Each skill includes a name, description (when to use it), step-by-step instructions, and optional references and assets. Skills follow the open Agent Skills specification.
Built-in skills
Port ships built-in skills for common platform workflows. These require no setup and are available to all users.
The troubleshoot-integration skill helps diagnose integration sync issues, mapping errors, and missing entities. It walks Port AI through identifying the integration, reviewing sync metrics and event logs, and recommending fixes in the correct format.
See the full list and behavior details in the Skills overview.
Custom skills in Port AI
You can create custom skills as entities in your software catalog. Custom skills override built-in skills when they share the same name, so your organization's process takes precedence.
To create a custom skill:
- Define the
skillblueprint in your data model. - Create skill entities with instructions, references, and assets.
- Ask Port AI or an agent to perform a task that matches the skill description.
For blueprint JSON, entity examples, and GitOps ingestion options, see Creating custom skills in the Skills overview.
Attach skills to agents
When you build an AI agent, you can configure which skills and tools the agent can access. Domain-specific agents combine catalog context with the right skills so automated runs follow your standards without manual prompting.
Next steps
- Skills overview: full reference for skill structure, creation, and best practices.
- Skills registry: publish skills company-wide with the Port CLI and skill groups.
- Port MCP Server: use skills from your IDE or external MCP clients.