Skip to main content
Glama

freshcheck

A dependency tree quietly ages. Some packages are a patch behind and can be bumped without a thought, some are a major version behind and need a planned upgrade, and some have not released anything in years — or their repository has been archived outright — and are a risk whatever version they are on. freshcheck takes the tree you have — each dependency's name, installed version, and latest available version, with optional abandonment signals — and sorts it into exactly those risk groups:

  • Behind level — for every dependency, current and latest are compared component by component, numerically. The first component that differs decides: major, minor, or patch. When latest is not ahead of current — equal, or even behind it — the level is none. Versions must be strict MAJOR.MINOR.PATCH semver: three numeric components, no leading zeros, no prerelease or build suffix.

  • Abandoned — a dependency counts as abandoned when its repository is archived, or when it has gone at least abandoned_days without a release (default 730, measured by the optional days_since_last_release). Abandonment outranks the version gap: an archived package two majors behind is grouped abandoned, with its behind level still reported alongside so the size of the gap is not lost.

  • Groups and counts — every dependency lands in one group — abandoned, major, minor, patch, or current (behind level none, not abandoned) — and the result carries the names in each group plus the group sizes and the overall total.

Every list in the result is sorted ascending, so the same tree always yields exactly the same assessment. Malformed input — a missing or empty dependency list, a missing, empty, or duplicate name, a version that is not strict semver, a wrongly typed signal — is a validation error naming the offending path (for example dependencies[1].latest must be a strict MAJOR.MINOR.PATCH semver version), and no partial assessment is produced. The computation is pure and local: no network, no registry lookups, no files read beyond the input you pass in — freshcheck judges only the versions and signals you supply.

CLI

npx @reneturcios/freshcheck assess --input args.json
npx @reneturcios/freshcheck --version

args.json contains the full arguments object:

{
  "dependencies": [
    { "name": "core", "current": "2.0.0", "latest": "2.0.0" },
    { "name": "lib", "current": "1.9.0", "latest": "1.10.2" },
    { "name": "old", "current": "0.8.1", "latest": "3.0.0", "archived": true }
  ]
}

archived defaults to false per dependency, and abandoned_days defaults to 730 for the whole call. The assessment is printed as JSON:

{
  "dependencies": [
    { "name": "core", "current": "2.0.0", "latest": "2.0.0", "behind": "none", "group": "current" },
    { "name": "lib", "current": "1.9.0", "latest": "1.10.2", "behind": "minor", "group": "minor" },
    { "name": "old", "current": "0.8.1", "latest": "3.0.0", "behind": "major", "group": "abandoned" }
  ],
  "groups": {
    "abandoned": ["old"],
    "major": [],
    "minor": ["lib"],
    "patch": [],
    "current": ["core"]
  },
  "counts": { "abandoned": 1, "major": 0, "minor": 1, "patch": 0, "current": 1, "total": 3 }
}

Note lib: 1.10.2 is a minor step ahead of 1.9.0 because components compare numerically, not lexicographically.

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: cf-package-check

MCP server

npx -y @reneturcios/freshcheck mcp

MCP client config:

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

Tools

  • assess_freshness — Assess how stale a dependency tree is, grouped by risk.

    • dependencies (array, required, at least one) — dependencies as {name, current, latest, archived?, days_since_last_release?} objects; name is unique across the set, and current / latest are strict MAJOR.MINOR.PATCH semver versions

    • abandoned_days (integer, optional, default 730, at least 1) — days without a release at which a dependency counts as abandoned

    Returns {dependencies, groups, counts}.

Docker

docker run -i ghcr.io/cloudcorprecords/freshcheck
docker run --rm -v "$PWD:/w" ghcr.io/cloudcorprecords/freshcheck assess --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
    A
    maintenance
    Translates a lockfile diff into a human-readable upgrade plan. For every dependency bump (npm or PyPI) returns semver class, breaking changes from GitHub release notes, CVEs fixed in the range, migration guide links, and a clear per-package recommendation. Bulk tool ranks up to 50 package changes in parallel by risk (security > caution > review > likely-safe > safe).
    2
    113 npm
    2
    MIT
  • A
    license
    Not graded
    quality
    C
    maintenance
    Enables agents to query whether a WordPress plugin, npm package, or PyPI project looks abandoned, returning a live abandonment-risk score, status, and the registry signals behind it, plus raw registry, GitHub, and version history detail. Calls are paid per request in USDC via x402, with free discovery and source-listing tools.
    MIT
  • A
    license
    A
    quality
    B
    maintenance
    Enables AI agents and developers to score npm packages, PyPI packages, and GitHub repositories for maintenance health, producing A–F grades, per-signal details, and plain-language risk lists before adoption. It supports checking individual packages, scanning manifests or lock files for transitive dependencies, and comparing alternatives.
    3
    109 npm
    Apache 2.0
  • A
    license
    Not graded
    quality
    B
    maintenance
    Exposes an assess_repo tool that scans a repository, local folder, or npm/PyPI package and returns supply-chain risk findings with file, line, and code evidence — install-time scripts, dangerous patterns, secrets, AI-agent manipulation, impersonation and published-artifact mismatches — while never executing the target's code. Findings carry severities from info to critical so results can gate installs and CI jobs.
    2
    Apache 2.0