Provides tools for inspecting and reviewing GitHub pull requests: listing open PRs in a repository, summarizing a PR's description, size, touched areas, review status and CI results, fetching changed files and diff hunks within a character budget, running rule-based risk checks (secrets in added lines, sensitive files, large diffs, missing tests, dependency/migration changes), and posting reviews with inline comments validated against the diff (including a dry-run preview).
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@deltaprobe review PR #42 in acme/app and flag any risky changes"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Give your AI assistant x-ray vision into pull requests.
deltaprobe is a Model Context Protocol (MCP) server that lets AI assistants like Claude, Cursor and VS Code Copilot probe GitHub pull requests: list them, read their diffs, summarize changes, flag risky patches and post review comments.
Available on npm and the official MCP Registry as io.github.Deepthi-04/deltaprobe.
Tool | Status | Description |
| ✅ | List open PRs in a repo, most recently updated first |
| ✅ | Overview of a PR: description, size, file types, most-touched areas, review status and CI results |
| ✅ | Fetch changed files and diff hunks, trimmed to whole hunks within a character budget |
| ✅ | Rule-based risk checks: secrets in added lines (masked), sensitive files, large diffs, missing tests, CI/migration/dependency changes, debug code |
| ✅ | Post a review with inline comments; every line is checked against the diff first, with a |
Related MCP server: GitHub PR Reviewer
You: "Review PR #42 in acme/app"
│
▼
AI assistant ──(MCP over stdio)──► deltaprobe ──► GitHub API
▲ │
└──── structured PR data ◄──────┘The assistant decides which deltaprobe tools to call, deltaprobe fetches and trims the data from GitHub, and the assistant writes the review.
Node.js 20+ (22.12+ for development)
A GitHub token (create one). Optional for reading public repos, but it raises the rate limit from 60 to 5,000 requests/hour. post_review_comment needs one with Pull requests: Read and write permission.
# public repos, read-only (no token)
claude mcp add deltaprobe -- npx -y deltaprobe
# with a token (higher rate limit, private repos, posting reviews)
claude mcp add deltaprobe -e GITHUB_TOKEN=<your-token> -- npx -y deltaprobeAdd to your MCP config:
{
"mcpServers": {
"deltaprobe": {
"command": "npx",
"args": ["-y", "deltaprobe"],
"env": { "GITHUB_TOKEN": "<your-token>" }
}
}
}Then ask: "What are the open PRs in modelcontextprotocol/typescript-sdk?"
git clone https://github.com/Deepthi-04/deltaprobe.git
cd deltaprobe
npm install
npm test # run unit tests
npm run build # compile TypeScript to dist/
npm run dev # run the server from sourcesrc/
├── index.ts # MCP server entry: registers tools, connects over stdio
├── github.ts # GitHub API client
├── diff.ts # hunk splitting, budget fitting, commentable lines
├── files.ts # file type and area classification
├── risk/rules.ts # risk rules: each takes the PR's files and returns findings
└── tools/ # one file per tool
test/ # vitest unit testslist_open_prs
get_pr_diff with smart truncation
summarize_pr
check_pr_risk rules
post_review_comment
CI on GitHub Actions
Publish to npm and the MCP Registry
List on Smithery
GitHub Action that reviews PRs automatically
Run rule-based risk checks on a pull request: possible secrets in added lines (masked), sensitive files, large diffs, source changes without tests, CI workflow, migration and dependency changes, and leftover debug code. Returns an overall level (none/low/medium/high) and findings with file and line where possible. These are heuristics: confirm each finding by reading the code with get_pr_diff before reporting it.
Inputs: repo (required), owner (required), pullNumber (required).
Get the changed files and unified diff hunks of a pull request. Patches are trimmed to whole hunks within a character budget; lock files and build output are skipped. Every file is listed, and files without a patch have an 'omitted' reason. If 'truncated' is true, call again with 'files' to read specific files in full.
Inputs: repo (required), owner (required), pullNumber (required), files, +1 more.
List open pull requests in a GitHub repository, most recently updated first. Use this to find which PR to review.
Inputs: repo (required), owner (required), limit.
Post a review on a pull request: an overall summary plus optional inline comments on specific lines. Inline comments must target lines in the PR's diff (RIGHT = new version, LEFT = removed lines); they're all checked first, and nothing is posted if any is invalid. Defaults to a neutral COMMENT review. Use dryRun to preview. Requires a GITHUB_TOKEN that can write to pull requests. Only post when the user has asked you to.
Inputs: body (required), repo (required), owner (required), pullNumber (required), +3 more.
Get an overview of a pull request without its diff: title, description, author, branches, labels, size, file types (source/test/docs/config/generated), most-touched areas, review status and CI check results. Call this first when reviewing a PR, then use get_pr_diff to read the code.
Inputs: repo (required), owner (required), pullNumber (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool targets a distinct stage of PR review: listing (list_open_prs), overview metadata (summarize_pr), code content (get_pr_diff), heuristic risk scanning (check_pr_risk), and writing feedback (post_review_comment). Descriptions explicitly state sequencing and warn against confusing heuristics with actual code reading, leaving no ambiguity.
All five names follow a consistent snake_case verb_noun pattern (list_, get_, summarize_, check_, post_) with a shared 'pr' domain noun. No mixed conventions or vague verbs.
Five tools map cleanly onto a coherent review workflow with no redundancy or filler. Each tool covers a necessary and non-overlapping step from discovery to posting a review.
The surface covers the core review lifecycle: discover PRs, understand them, read diffs, assess risk, and post comments with inline feedback. Minor gaps remain around reading existing review threads/replies and retrieving CI log details, but agents can work around these.
Operate GitHub from your AI assistant: repos, PRs, issues, Actions. Previews and read-only mode.
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
Access the GitHub API, enabling file operations, repository management, search functionality, and…
Read and write KukGit repositories, files, issues and pull requests from an AI assistant.
