Skip to main content
Glama

sizecheck

A pull request can be three lines in one file or three thousand lines spread across half the repository, and the review burden is not the same. sizecheck takes the list of changed files — each file's path, the lines it adds and deletes, and optionally how many of its exported symbols changed — and turns it into one blast-radius report:

  • Score and risk — a single integer score: total churn (lines added plus lines deleted), plus 25 points for every exported symbol touched, plus 10 points for every extra top-level directory beyond the first. The score maps to a risk band: under 100 is small, 100–399 is medium, 400–999 is large, and 1000 or more is huge.

  • Totals — the file count and the summed lines added, lines deleted, churn, and exports changed, so the score can always be checked against its inputs.

  • Per-directory spread — the change grouped by top-level directory (a file with no directory in its path counts under "(root)"), sorted by churn descending and then directory name ascending.

  • Hottest files — the five highest-churn files, ties broken by path ascending, so the files that deserve the closest read surface first.

  • Split suggestions — when the risk is large or huge, every directory group except the largest is listed as a part of the change that could land separately. Small and medium changes, and changes confined to a single directory, get no suggestions.

Every changed file needs a unique, non-empty path and non-negative integer added and deleted counts. exports_changed is optional (default 0) and must be a non-negative integer when given; kind is optional (default "code") and must be one of "code", "test", "docs", "config", or "other" when given — it is validated but does not affect the score. Invalid input is a validation error naming the offending path (for example files[0].added must be an integer >= 0) and no partial report is produced. The computation is pure and local: no network, no configuration, no files read beyond the input you pass in.

CLI

npx @reneturcios/sizecheck score --input args.json
npx @reneturcios/sizecheck --version

args.json contains the full arguments object:

{
  "files": [
    { "path": "src/api.ts", "added": 120, "deleted": 30, "exports_changed": 3 },
    { "path": "src/api.test.ts", "added": 80, "deleted": 10, "kind": "test" },
    { "path": "docs/api.md", "added": 25, "deleted": 5, "kind": "docs" }
  ]
}

The report is printed as JSON:

{
  "score": 355,
  "risk": "medium",
  "files_count": 3,
  "added": 225,
  "deleted": 45,
  "churn": 270,
  "exports_changed": 3,
  "by_directory": [
    { "directory": "src", "files": 2, "churn": 240 },
    { "directory": "docs", "files": 1, "churn": 30 }
  ],
  "hottest_files": [
    { "path": "src/api.ts", "churn": 150 },
    { "path": "src/api.test.ts", "churn": 90 },
    { "path": "docs/api.md", "churn": 30 }
  ],
  "split_suggestions": []
}

The score is 270 churn + 25 × 3 exports + 10 × 1 extra directory = 355.

Exit codes: 0 on success, 1 on a runtime error (including an unreadable or invalid input file, or invalid input), 2 on a usage error.

Related MCP server: code-intel

MCP server

npx -y @reneturcios/sizecheck mcp

MCP client config:

{
  "mcpServers": {
    "sizecheck": {
      "command": "npx",
      "args": ["-y", "@reneturcios/sizecheck", "mcp"]
    }
  }
}

Tools

  • score_blast_radius — Score a pull request's blast radius from its changed files.

    • files (array, required, at least one) — files as {path, added, deleted, exports_changed?, kind?} objects; path is unique across the set

    Returns {score, risk, files_count, added, deleted, churn, exports_changed, by_directory, hottest_files, split_suggestions}.

Docker

docker run -i ghcr.io/cloudcorprecords/sizecheck
docker run --rm -v "$PWD:/w" ghcr.io/cloudcorprecords/sizecheck score --input /w/args.json

Development

npm install
npm run build   # tsc -> dist/
npm test        # tsx --test test/*.test.ts (pure local, no network)
npm run dev     # run the CLI from source

License

MIT — Copyright (c) 2026 Rene Turcios

Related MCP Connectors

Related MCP Servers

  • A
    license
    A
    quality
    C
    maintenance
    Exposes the squad-dev workflow as deterministic MCP tools, enabling task classification, risk scoring, specialist reviewer selection, and advisory verdict consolidation for code changes.
    27
    22 npm
    4
    Apache 2.0
  • A
    license
    Not graded
    quality
    A
    maintenance
    MCP server for persistent code exploration and diff-aware review, exposing explore and review tools to map call paths, blast radius, and generate explainable risk-scored reports.
    12 npm
    2
    MIT
  • F
    license
    A
    quality
    C
    maintenance
    Exposes Semgrep as deterministic, rule-matching analysis tools over MCP so a code-review agent can scan only the lines changed in a diff, scan whole files on demand, and check analyzer availability, version, ruleset, and limits. Results are returned as compact per-finding lines with source snippets suitable as comment anchors, along with counts of suppressed out-of-diff matches.
    3
    -