Rovo Skill (EAP)

The rovo:skill module packages reusable instructions that help Rovo Agents complete specialized tasks. A skill consists of a SKILL.md instruction file, optional supporting files, and optional action dependencies from the same Forge app.

Forge validates and bundles each skill when you deploy the app. During the EAP, you can deploy apps that use rovo:skill only to development environments.

Manifest structure

1
2
3
4
5
6
7
8
9
modules {}
└─ rovo:skill []
   ├─ key (string) [Mandatory]
   ├─ source {} [Mandatory]
   │  └─ dir (string) [Mandatory]
   └─ dependencies {} [Optional]
      └─ tools [] [Mandatory if dependencies is specified]
         └─ action (string)

Properties

Skill directory

Each skill has its own directory. At minimum, the directory must contain a SKILL.md file:

1
2
3
4
5
skills/
└─ jira-issue-analyst/
   ├─ SKILL.md
   └─ references/ [Optional]

The directory can contain supporting reference documents. Refer to supporting files from SKILL.md using paths relative to the skill directory, for example references/issue-fields.md.

SKILL.md format

The SKILL.md file follows the Agent Skills specification. It must contain YAML frontmatter followed by Markdown instructions:

1
2
3
4
5
6
7
8
9
10
11
12
---
name: jira-issue-analyst
description: Retrieve and summarize Jira issues for a project or sprint when users need an issue analysis.
allowed-tools: get-project-issues
---

# Jira issue analysis

Use `get-project-issues` to retrieve the issues for the project or sprint requested by the user.

Summarize the result by status and priority. Highlight blocked work and issues without an assignee.

The name must match the name of the directory that contains SKILL.md. In this example, the directory is skills/jira-issue-analyst.

The following frontmatter fields are supported:

Write the Markdown body as instructions for the Agent. Explain when and how to use every action in dependencies.tools, how to interpret its results, and how to handle expected errors or edge cases. Keep the Markdown body of SKILL.md to 500 lines or fewer. Forge CLI warns when the body exceeds this length. Move detailed material into focused files under references.

Manifest example

The following example shows only the relevant modules. A complete manifest also requires an app section that defines the app ID and runtime, and a permissions section that declares the scopes required by the actions.

It defines a skill backed by an action and makes that skill available to a Forge Rovo Agent:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
modules:
  function:
    - key: get-project-issues-function
      handler: src/index.getProjectIssues

  action:
    - key: get-project-issues
      name: Get project issues
      function: get-project-issues-function
      actionVerb: GET
      description: Retrieve Jira issues for a project
      inputs:
        projectKey:
          title: Project key
          type: string
          required: true
          description: The key of the Jira project to retrieve issues from

  rovo:skill:
    - key: jira-issue-analyst
      source:
        dir: skills/jira-issue-analyst
      dependencies:
        tools:
          - get-project-issues

  rovo:agent:
    - key: issue-analyst-agent
      name: Issue analyst
      description: An Agent that analyzes work in a Jira project
      prompt: Help users understand the status of work in their Jira projects.
      skills:
        - jira-issue-analyst

Declare the OAuth scopes required by the actions in the app's permissions.scopes. The rovo:skill module doesn't introduce additional scopes or a separate consent step.

Use a skill with a Forge Rovo Agent

To make a skill available to a Forge Rovo Agent, add the rovo:skill module key to the Agent's skills property. A Forge Rovo Agent can access only the skills declared in its skills property.

At runtime, the Agent selects a skill based on its description and the user's request. Explicit invocation by name isn't supported during the EAP.

Limits and restrictions

• A skill directory must not exceed 100 MB uncompressed.
• Tool dependencies must be action modules declared by the same app.

EAP limitations

• Apps that declare rovo:skill can be deployed only to a development environment.
• Skill-to-skill dependencies aren't supported.
• Executable skill sources, including scripts, aren't supported.

Related links

• Agent Skills specification
• Action module
• Rovo Agent module