speckit-extensions
A Claude Code plugin for AI-assisted spec-driven development. Provides skills that gather context from external systems such as Jira, Confluence, Figma, and others, and pipes it into speckit commands to drive spec driven development workflows.
Installation
1. Add the marketplace
claude plugin marketplace add yesmarket/claude-marketplace
This adds the yesmarket/claude-marketplace as a source for plugin installation.
2. Install the plugin
claude plugin install speckit-extensions@yesmarket/claude-marketplace
This plugin bundles the following MCP servers directly — no separate dependency installation required:
Why bundle MCP servers directly? At the time of writing, Claude Code does not auto-install plugin dependencies, so declaring an Atlassian plugin dependency would require users to install it manually. Bundling the MCP server directly keeps installation to a single step. If plugin dependency auto-installation is supported in a future release, this plugin may switch to declaring dependencies instead.
Note: If you also have the official Atlassian plugin installed, there are no conflicts. Claude Code handles duplicate MCP server registrations gracefully.
Skills
constitution-from-confluence
Fetches a Confluence page and its full child hierarchy, feeding each page individually into /speckit.constitution.
/speckit-extensions:constitution-from-confluence <confluence-url> [additional-input]
Arguments:
| Argument | Required | Description |
|---|
confluence-url | Yes | URL to a Confluence page, e.g. https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/Page+Title |
additional-input | No | Extra context or instructions passed as a separate invocation of /speckit.constitution |
Example:
/speckit-extensions:constitution-from-confluence https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/Home
/speckit-extensions:constitution-from-confluence https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/Home focus on the authentication domain
The skill parses the space key and page ID from the URL, recursively fetches all descendant pages depth-first via the Atlassian MCP server, and invokes /speckit.constitution once per page. If additional input is provided, it is passed as a final separate invocation.
Authentication: The Atlassian MCP server requires authentication with your Atlassian account.
specify-from-jira
Fetches a Jira ticket and uses it as context for /speckit.specify.
/speckit-extensions:specify-from-jira <jira-ticket-id> [additional-input]
Arguments:
| Argument | Required | Description |
|---|
jira-ticket-id | Yes | Jira issue key, e.g. PROJ-123 |
additional-input | No | Extra context or instructions to include alongside the ticket |
Example:
/speckit-extensions:specify-from-jira PROJ-123
/speckit-extensions:specify-from-jira PROJ-123 focus on the mobile experience
plan-from-confluence
Fetches a Confluence page and its full child hierarchy, feeding each page individually into /speckit.plan.
/speckit-extensions:plan-from-confluence <confluence-url> [additional-input]
Arguments:
| Argument | Required | Description |
|---|
confluence-url | Yes | URL to a Confluence page, e.g. https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/Page+Title |
additional-input | No | Extra context or instructions passed as a separate invocation of /speckit.plan |
Example:
/speckit-extensions:plan-from-confluence https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/HLD
/speckit-extensions:plan-from-confluence https://mycompany.atlassian.net/wiki/spaces/PROJ/pages/123456789/HLD focus on the data access layer
The skill parses the space key and page ID from the URL, recursively fetches all descendant pages depth-first via the Atlassian MCP server, and invokes /speckit.plan once per page. If additional input is provided, it is passed as a final separate invocation.
Authentication: The Atlassian MCP server requires authentication with your Atlassian account.
plan-from-lucid
Fetches a Lucidchart diagram page and uses it as context for /speckit.plan.
/speckit-extensions:plan-from-lucid <lucid-url> [additional-input]
Arguments:
| Argument | Required | Description |
|---|
lucid-url | Yes | URL to a Lucidchart document page, including the page query parameter |
additional-input | No | Extra context or instructions to include alongside the diagram |
Example:
ChromeDevTools
affaan-m
f
anthropics
1270011
figma