Skip to main content
Key characteristics:
  • Reusable: Write once and attach to as many agents as needed. Updates propagate automatically to every agent using the Skill.
  • Named and invocable: Each Skill has a unique name and can be triggered mid-conversation by typing ~skill-name in the chat composer.
  • Versioned: Maintain multiple named versions alongside the always-present base version, so you can iterate safely without breaking existing agents.
  • Modular: Combine multiple Skills on a single agent to compose complex behavior from small, independently tested pieces.

​
Skills in ELITEA

ELITEA Skills let you encapsulate a specific capability — a focused set of Markdown instructions — and share it across agents without copying text. When an agent activates a Skill, its instructions are injected into the conversation for that turn, giving the agent the targeted behavior defined in the Skill. What makes Skills powerful:
  • Single source of truth: Update the Skill once; all attached agents immediately reflect the change.
  • Tested in isolation: Each Skill has a built-in test chat panel so you can verify behavior before attaching it to any agent.
  • Project-scoped sharing: Any agent in the same project can attach and use the same Skill.
  • Attach up to 5 per agent: Combine up to five Skills on one agent version to build layered, modular capabilities.
Perfect for shared reasoning templates, persona or tone guidance, domain-specific procedures, and any behavior that multiple agents need to perform consistently.

​
Example: Pull Request Review Skill

Scenario: Maria is a tech lead whose team uses three different agents — a code review agent, a documentation agent, and an onboarding agent. All three need to evaluate pull request descriptions using the same quality checklist. Instead of copying the checklist into each agent, she creates a single Skill. Maria’s Task: “I need all three agents to check that every PR has a clear summary, lists affected components, includes test coverage notes, and flags missing ticket references.” Skill Configuration:
  • Name: pr-review
  • Description: Evaluates pull request descriptions for completeness and quality.
  • Instructions:
How the Skill Works:
  1. Maria creates the Skill once from the Skills menu and saves it.
  2. She opens each of the three agents and attaches pr-review from the Skills section.
  3. During a conversation, any team member can type ~pr-review to invoke the Skill, or the agent can activate it automatically based on its instructions.
  4. The agent applies the checklist to the PR description and returns a structured review.
Sample Output:
PR #412 Description Review:
  • ✅ Present: Summary — “Adds retry logic to the payment service”
  • ✅ Present: Affected components — payment-service/, shared/http-client
  • ❌ Missing: Test coverage — Add a note describing which test cases were added or updated
  • ❌ Missing: Ticket reference — Link the related Jira ticket (e.g., PAY-301)
When Maria updates the checklist in the Skill, all three agents pick up the change automatically — no edits needed on each agent individually.
Ready to get started? Explore the Skills menu to create your first Skill.For detailed definitions of terms and concepts, please refer to the ELITEA Glossary