Keep a personal website's content updated automatically
Posted on Jan 14. Task: The author automated adding podcast episodes to a Nuxt content site. They used Goose with Playwright MCP to scrape metadata and images, Cloudinary MCP to upload assets, and GitHub MCP to create pull requests. The human role involved configuring the tools, providing the podcast URL, and reviewing the generated code before merging.; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: The setup required configuring MCP servers for Playwright, Cloudinary, and GitHub, including API keys. The author noted it took time to figure out the process initially. Quality drift in batch processing was mentioned as a potential concern in comments, though the author stated they had not tested with endless amounts of URLs..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
Posted on Jan 14
Workflow
The author automated adding podcast episodes to a Nuxt content site. They used Goose with Playwright MCP to scrape metadata and images, Cloudinary MCP to upload assets, and GitHub MCP to create pull requests. The human role involved configuring the tools, providing the podcast URL, and reviewing the generated code before merging.
Reported result
The workflow saved significant time and allowed bulk updates via multiple URLs. The author confirmed images appeared in Cloudinary and the site rendered correctly after merging the pull request. No specific metrics on time saved were provided, but the process was described as working well for single and batch inputs.
Human role
The original post does not say how a person stayed involved.
Limitations
The setup required configuring MCP servers for Playwright, Cloudinary, and GitHub, including API keys. The author noted it took time to figure out the process initially. Quality drift in batch processing was mentioned as a potential concern in comments, though the author stated they had not tested with endless amounts of URLs.
Try it
Act as an automation engineer. I need to set up a workflow to add podcast episodes to my [replaceable: website tech stack, e.g., Nuxt/Next.js] site. 1. **Setup**: Configure Goose (or similar agent) with Playwright MCP, Cloudinary MCP, and GitHub MCP. Ensure API keys for Cloudinary and GitHub are stored securely in the agent's settings. 2. **Input**: I will provide a [replaceable: podcast episode URL]. 3. **Execution**: - Use Playwright MCP to navigate to the URL and extract metadata (title, date, host, description) and the episode image. - Use Cloudinary MCP to upload the image to my [replaceable: Cloudinary folder] and retrieve the public ID. - Generate a new Markdown file with YAML frontmatter containing the extracted data and Cloudinary image URL. - Use GitHub MCP to create a new branch, commit the file, and open a Pull Request. 4. **Verification**: Before merging, run a local dev server and use Playwright to screenshot the new page to verify rendering. 5. **Human Check**: Present the PR link and screenshot to me. Do not merge until I explicitly approve. If any MCP fails or metadata is missing, stop and report the error. Start with a dry run on one URL.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.