To find out what changed in the codebase this week without reading a diff, ask Claude. With your repositories connected to DeployIt, the answer lists what landed on the tracked branch, grouped into deliveries, each with its date, in plain words and with no name.
This page is for product managers who prepare a team update, and for support leads who want to warn customers before the tickets arrive. It shows the answer, what the tool behind it returns, and the questions that follow.
What changed in the product this week?
Three deliveries since Monday.
- Wednesday, fix: a webhook whose signature has expired is now rejected.
- Thursday, feature: attachments are limited to 20 MB by default.
- Friday, feature: in the CSV export, the amounts column has a new name and comes last.
Why is "what changed in the codebase this week?" so hard to answer?
Everything that changed in your product is written down, to the character. Every fix, every new default, every modified behaviour left an exact trace: a diff, dated, tied to a commit and often to a pull request.
The information is not missing. It is unreadable for almost everyone: reading diffs takes repository access, technical context and time. So support interrupts a developer, the product manager rereads notes written for the previous release, and customer success answers from memory. Every written summary is a copy made on one date, and copies drift.
Example: a ticket lands on Monday morning, "the amounts column is no longer where it was in the CSV export". The right first question is not "who is free to look at this?". It is "what changed in the export since Friday?". When that question is answered at once, the ticket is qualified and nobody was interrupted.
What do recent_changes and what_changed return?
Two entries, both described on the docs page. You do not have to call them: Claude chooses.
| What you give | What comes back | |
|---|---|---|
recent_changes (tool) | A number of days, from 1 to 90, 14 by default; a repository if needed | What landed on the tracked branch, grouped into deliveries (by pull request, or by day, module and type): date, type, title, modules, number of commits. No person's name. Immediate |
what_changed (prompt) | A number of days, optional, 7 by default | What changed in the product: a ready-made request, so the summary comes back in the same form every week |
Both are open to the Everyone, Developers and Leadership profiles. What you read is Claude's wording of that list: the deliveries, in the words of someone who uses the product.
Does one delivery raise a question? Ask it. "What does the export do with the amounts column now?" goes to ask_code, which investigates in the code and gives its answer a status: Verified, Backend only, Not found or Inconclusive.
Who asks, and what for?
"What changed this week?" is the most common form of a family of questions:
- Product, before the weekly update: "What changed this week in checkout?"
- Support, in front of a ticket: "Did anything change in the export since Friday?" The method is in how to reduce developer interruptions from support.
- Customer success, before a call: "What changed this month in billing?"
- Whoever answers an integrator: "Did the API change this week? Does anything break?"
That last one deserves care. The list tells you that a delivery touched the API, and when. Whether an integration breaks is a question about behaviour: ask "does the invoices endpoint still accept the old parameter name?" and read the status before you write to the integrator.
A feature shipped and nobody was told: how do you catch it?
It happens in every team. Something lands, the announcement does not, and support discovers the feature when a customer mentions it. The list of deliveries catches it, because it is built from what landed, not from what was announced.
Then ask how the feature behaves. If the answer comes back Backend only, the product can do it but no screen leads there: do not send a customer looking for a button.
Why did this behaviour change?
Two questions hide in that one. "Since when, and in which delivery?" is in the list: a date, a type and a title, and the title often says what the change was for. "What does the product do now?" goes to ask_code.
The reason behind a decision is not always written anywhere. When neither the title nor the code says it, ask the person who knows, once, with the date in hand.
Back from leave: what changed over a month?
The window goes up to 90 days. "What changed over the last 30 days?" works the same way: it is the question to ask after three weeks away, or when you take over an account.
A long window returns many deliveries. Narrow it to one part of the product ("in billing") or to one repository. Beyond 90 days the question changes: ask how the product behaves today.
Can't I just connect my repository to Claude?
Claude's own GitHub connector syncs the files of one branch: not the commit history, not the pull requests, according to Claude's documentation (checked on 17 November 2026). Claude then sees the code as it is, not what moved. What that connector sees, and what it cannot has its own page.
The official GitHub MCP server does expose commits. Its list_commits tool filters by date and its get_commit tool returns one commit, with the full patch on request; no tool groups a week into deliveries (its README, checked on 17 November 2026). A developer gets real value from it. It also asks each person for a GitHub account or a GitHub token, which your support and product teams have no reason to hold: the detail is here.
What DeployIt does here
DeployIt is an MCP server. Your company gets one link, https://mcp.deployit.ai/<company>/<code>; each person adds it to Claude (on the web, on the desktop or in Claude Code) and signs up with a name and an email, which is not verified; their questions are logged under that name. Everyone starts with the Everyone profile: questions about the product, answered from the code; a filter removes the code, the file paths and the names of team members it recognises from the answers, without proving their absence, and a leak test tells you what still gets through. An administrator gives the Developers profile (excerpts) or the Leadership profile (developer statistics) to whoever needs it. Repositories are connected from GitHub or GitLab; DeployIt makes no commit, branch or pull request, and only adds one webhook per repository.
That webhook tells DeployIt about new commits, and the server_info tool says how fresh its reading is. recent_changes names no one. Who shipped what is a separate question, for the Leadership profile: DeployIt tracks each developer, and the company informs the people it tracks.
DeployIt publishes nothing: no page, no email. The answer goes to the person who asked, in Claude. The rest of the product (profiles, developer statistics, price) is in DeployIt: code questions and developer statistics in Claude, and the product team's example is on the home page.
Next step: add the demo address below to your Claude connectors and ask "What changed this week?"; then connect your repositories, add your company link to Claude in three steps and ask the same question about your own product.
Price. Free plan: 15 questions a week for the whole company. A 30-day trial, no card. Plans from €290 a month excluding tax for up to 25 tracked developers: see pricing.
Ask your code from Claude
30-day trial, no card. Or try the demo first: add https://mcp.deployit.ai/demo to your Claude connectors.
Frequently asked questions
How do I see what changed in the codebase this week without reading diffs?
Ask Claude, once your repositories are connected to DeployIt. The recent_changes tool returns what landed on the tracked branch, grouped into deliveries, each with its date, type, title, modules and number of commits. Claude puts that list into plain words; no code is shown and no person is named.
Can I ask what changed over the last month, or since I went on leave?
Yes: the window goes from 1 to 90 days, and it is 14 days when you give none. A long window returns many deliveries, so narrow the question to one part of the product or one repository. Beyond 90 days, ask how the product behaves today instead.
Does the answer say which developer made each change?
It never does. What changed is described without any person's name, whatever your profile. Who shipped what is a separate question, reserved for the Leadership profile, which reads the statistics of each developer.
Is this a changelog or a release notes generator?
No. DeployIt generates no changelog, no release notes and no page: it answers a person who asks, in Claude. What you tell your customers about a release stays yours to write; the answer gives you what really landed to start from.
Why can Claude not tell me what changed with its own GitHub connector?
Because that connector syncs the files of one branch, not the commit history or the pull requests, according to Claude's documentation (checked on 17 November 2026). Claude then sees the code as it is, not what moved. A list of what landed needs a source that reads the history.
Continue reading
How to connect Claude to a private GitHub repo without sharing the code
For a CTO: two ways to connect Claude to a private GitHub repo. Claude's own connector hands the files to whoever connects, with that person's GitHub access; one company link gives each person a profile and, with the Everyone profile, answers from which a filter removes the code (filtered and tested). The steps, in order.
MCP server security testing: how a leak test works, and how to read it
For a security lead or a CTO: most MCP server security testing checks who gets in. A leak test checks what gets out: trap questions fired at each profile that must not see code, a sorted report, and a warning that blocks nothing.
DeployIt: ask your code from Claude, and follow each developer's work
For the person who decides, CTO, CEO or head of product: the whole of DeployIt in one read. One link for your company's Claude, answers from the code without giving access to it, statistics on each developer, what happens to your code, and the price.