Fern Writer is a Slack-based technical writing agent that keeps your docs aligned as your product evolves. It’s powered by Devin from Cognition. Fern Writer understands Fern components and your writing style, and can be customized via an AGENTS.md file in your docs repository.
In Slack channels where you’ve added Fern Writer, tag @Fern Writer and describe the change you need. Fern Writer only responds when directly tagged. It will react to your message to confirm receipt, then create a pull request and reply with a link.
Fern Writer supports image and file attachments for additional context. When tagged in an existing thread, it reads the full conversation to understand context before responding.
Request changes by commenting in the Slack thread. Once the PR meets your requirements, merge it like any other pull request.
Each pull request includes a Requested by field in the description, attributing the change to the person or team that initiated the request. Commits are signed and attributed to fern-support, ensuring that automated changes are clearly distinguishable from manual contributions in your repository’s history.
@Fern Writer document the new rate limiting feature added in PR #123@Fern Writer add a section about webhook retry behavior to the webhooks guide@Fern Writer merge the authentication and authorization pages, and add a redirect from the old auth page@Fern Writer fix the broken code example in the quickstart and update the package version to v2.1.0Fern Writer only supports GitHub. GitLab and other Git providers aren’t supported.
To start using Fern Writer, add it to your Slack workspace (you must be a Slack admin) and invite it to channels where your team discusses documentation.
Get a unique Slack installation link for your organization. Provide:
owner/repo format (e.g., acme/docs). The Fern GitHub App must be installed in the repository.You can alternatively use this cURL request:
Follow the URL returned in the response.
Fern Writer doesn’t store your Slack messages directly. When you tag @Fern Writer or reference a message or thread, the content is stored in a session to generate the documentation pull request. Session data isn’t retained after the task completes.
Neither Fern nor Devin uses your data to train AI models. Fern explicitly configures its Devin integration to opt out of any data collection for model training. Your channel messages, code, and documentation content are never used for training purposes. See Devin’s documentation on Slack integration security for additional details.