Orchestrates setup, testing, and teardown of the jira-mcp-direct recipe under agy-recipes/. Use when the user requests deployment, replication, or testing of the two direct Atlassian Jira MCP connectors (Rovo hosted + Custom Cloud Run).
Scanned 6/12/2026
Install to Claude Code
npx -y skills add jchavezar/vertex-ai-samples --skill replicating-jira-mcp-direct --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Replicating Jira Mcp Direct?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/jchavezar-replicating-jira-mcp-direct)More formats (shields.io, HTML) on the badges page.
---
name: replicating-jira-mcp-direct
description: Orchestrates setup, testing, and teardown of the jira-mcp-direct recipe under agy-recipes/. Use when the user requests deployment, replication, or testing of the two direct Atlassian Jira MCP connectors (Rovo hosted + Custom Cloud Run).
---
# Replicating Jira MCP Direct Integration Recipe
This skill guides the agent and user through deploying, validating, and destroying the direct Atlassian Jira MCP integration connectors in Gemini Enterprise.
## 📋 Pre-Flight Checklist
1. Ensure a `.env` file exists in `agy-recipes/jira-mcp-direct/` with:
* `GE_PROJECT_ID` (GCP Project ID, e.g. `vtxdemos`)
* `GE_PROJECT_NUMBER` (GCP Project Number, e.g. `254356041555`)
* `GE_ENGINE_ID` (Gemini Enterprise Engine ID, e.g. `jira-testing_1778158449701`)
* `ATLASSIAN_CLIENT_ID` (Standard Atlassian App Client ID, for Option C)
* `ATLASSIAN_CLIENT_SECRET` (Standard Atlassian App Client Secret, for Option C)
2. Ensure you have active `gcloud` credentials authorized for your target project.
---
## 🚀 Setup & Deploy
Execute the setup script:
```bash
uv run agy-recipes/jira-mcp-direct/scripts/setup.py
```
### Required Manual Console Steps (OAuth Consent)
Because Atlassian requires interactive user authorization (3LO), you must complete the final wiring in the Cloud Console:
1. **Atlassian Hosted Rovo Connector (`jiramcp-rovo-recipe`)**:
* Open the GE Console -> **Data stores** -> select `jiramcp-rovo-recipe_mcp_data` -> **Actions** tab.
* Click **Reload custom actions** to sync tools.
* Enable desired tools (e.g. `searchJiraIssuesUsingJql`, `getJiraIssue`).
* Click **Re-authenticate** and paste DCR Client ID/Secret from `~/.secrets/atlassian-rovo-dcr-ge.json`.
* Complete the Atlassian consent flow for your Jira site (Jira only, uncheck Confluence/Compass).
2. **Custom Cloud Run MCP Connector (`jiramcp-custom-recipe`)**:
* Open the GE Console -> **Data stores** -> select `jiramcp-custom-recipe_mcp_data` -> **Actions** tab.
* Click **Reload custom actions** to sync tools.
* Enable desired tools and click **Re-authenticate**.
* Paste your standard Atlassian App Client ID and Secret, and complete the consent flow.
---
## 🧪 E2E Verification
Once both connectors are active (showing green checks in the console), run the test script:
```bash
uv run agy-recipes/jira-mcp-direct/scripts/test_recipe.py
```
---
## 🧹 Cleanup & Teardown
To delete all resources created by this recipe:
```bash
uv run agy-recipes/jira-mcp-direct/scripts/teardown.py
```
This cleanly removes both datastores in Gemini Enterprise, detaches them from the engine, and deletes the Cloud Run server.
No comments yet. Be the first to comment!