Your changelog
was a chore. Now it ships itself.
The AnnounceKit MCP server puts your changelog inside Claude, ChatGPT, and every AI agent your team runs. Ask your agent to announce what shipped in Linear, Jira, or GitHub - it drafts, targets, and publishes for you, or saves a draft for your review. Don't make customers wait for updates you already shipped.
trusted by 2,000+ teams · 15-day trial · no credit card
Ship → Agent → Announced.
While you're still in the standup.
Your agents already know what shipped. With MCP they can now tell your customers: drafting the post, picking the segment, and publishing to every channel.
Your AI agents, wired to AnnounceKit
The agent reads the diff, writes two draft styles, applies your tone rules, targets the right segment, and calls publish_post.
From "Done" to announced
in under a minute.
Work finishes
Your agent notices the Linear issue moved to Done. No webhook glue, no Zap to babysit: it's a native MCP tool call away.
The agent drafts
It reads the issue, the diff, and your past posts. Then it writes the announcement in your voice, categorized and labeled.
Targeted, not blasted
Enterprise-only fix? It picks the enterprise segment. Public feature? Everyone. Your segmentation rules apply automatically.
Published everywhere
One post fans out to your changelog, in-app widget, email, Slack, and RSS. Your customers knew before your standup ended.
Six things to try
in your first session.
Every one of these maps to a real MCP tool. Copy a prompt, paste it into Claude, ChatGPT, or Cursor, and watch it happen.
Announce what shipped
"Announce what we shipped in ANN-1421. Match the tone of our last posts."
The agent reads your recent posts, drafts the announcement, and publishes or saves it for review.
Line up a launch post
"Draft the v3.2 announcement and schedule it for Tuesday 9am."
Posts can be drafted, updated, and scheduled - your launch goes out on time, every channel at once.
Close the loop on requests
"Summarize this week's feedback, then reply to the top feature request with our plan."
The agent reads feedback, and creates, comments on, or replies to feature requests for you.
Keep the roadmap honest
"Add SSO improvements to the roadmap and mark the export feature as shipped."
Roadmap items and statuses are one tool call away - no dashboard tab needed.
Ask how it landed
"How did last week's posts perform? What is our NPS trend saying?"
Post stats and NPS are readable over MCP, so your agent can report on reach and sentiment.
Draft the week in review
"Draft a weekly digest post from everything we published since Monday."
The agent lists and reads your posts, then drafts the roundup - you approve and publish.
Safe by design: the MCP server can read, create, update, publish, and schedule - it can never delete anything or touch billing. Prefer a human in the loop? Have the agent save drafts and publish from the dashboard. ChatGPT remote connectors require Developer Mode. Full capabilities in the MCP docs.
Automation on top.
The whole communication stack below.
Agents can do everything you can
Create, list, and publish posts. Read feedback and feature requests. Check NPS. Draft with your templates. If it's in the dashboard, it's a tool call.
Auto-publish or approve first
Let agents publish directly, or keep every AI draft in review until a human clicks Publish. Your workflow rules decide, per project.
A branded home for every update
Hosted on your own domain, SEO-friendly, multi-language, searchable. The source of truth your agents write to.
10+ ways to reach users in-app
Badge, sidebar, popup, toast, embed. Users see what changed without leaving your product.
Requests and NPS close the circle
Users vote, you ship, the agent announces, satisfaction gets measured. The loop runs itself.
Linear, Jira, GitHub + 7,000 more
Native integrations plus Zapier and a full GraphQL API with 11 webhook events for anything custom.
The last changelog post
you'll ever write by hand.
Connect the MCP server to your agent in two minutes. 15-day free trial, every feature included, no credit card.