Skip to content
modelcontextprotocol/typescript-sdk (opens in a new tab)

Solid SDK with excellent docs and clear governance, but pre-alpha status and PR restrictions create friction for newcomers wanting to contribute.

TypeScriptβ˜… 12,670450 open issuesOther
Overall score: 72 out of 100
Holding

last scored 2026-06-15

Readiness lenses

The same repo, two ways: how ready it is for an AI coding agent, and for a human newcomer. Switch lenses to see each one's score and checks.

How ready the repo is for an AI coding agent to read, build, and operate it.

24 of 35 checks passed across 6 categories.

  • Machine Orientation

    4 checks passed4 problems found

    • lowAgent instructions stay lean

      A bloated instructions file burns context and measurably hurts more than it helps.

      Fix: Trim AGENTS.md to the load-bearing commands and conventions; link out to the rest.

    • lowllms.txt present

      An llms.txt is the emerging machine-orientation standard β€” a curated map for LLMs.

      Fix: Add an llms.txt pointing at the key docs and entry points.

    • lowTopics set

      GitHub topics situate the repo's domain and stack for an agent at a glance.

      Fix: Add a few GitHub topics describing the domain and stack.

    • lowArchitecture doc

      An ARCHITECTURE doc hands an agent the module map instead of making it infer one.

      Fix: Add an ARCHITECTURE.md describing the major components and how they fit.

    • README present β€” An agent reconstructs what the repo is and how to work in it from the README first.
    • Agent instructions file β€” AGENTS.md / CLAUDE.md is the direct, machine-read channel for entry points and conventions.
    • Repository description set β€” The one-line description is the fastest orientation signal an agent reads.
    • Contribution conventions are written down β€” A contributor guide is where an agent learns the branch, commit, and review conventions it must follow.
  • Reproducible Setup

    3 checks passed3 problems found

    • mediumContainer or dev-container

      A Dockerfile / compose / devcontainer removes 'works on my machine' from setup.

      Fix: Add a Dockerfile, a compose file, or a .devcontainer so the environment is declared, not assumed.

    • lowEnv template present

      A .env.example tells an agent which configuration keys it must supply to run.

      Fix: Ship a .env.example listing every required key with placeholder values.

    • lowDev container

      A .devcontainer gives an agent a one-command, fully-declared environment.

      Fix: Add a .devcontainer so the toolchain is declared, not assumed.

    • Lockfile committed β€” Without a lockfile two clones can resolve different dependency trees β€” the build isn't reproducible.
    • Runtime / toolchain pinned β€” An unpinned runtime makes an agent guess the version a clean environment needs.
    • Version pins are exact β€” A floating pin (a bare major, or a range) still drifts between installs.
  • Code Navigability

    1 check passed1 problem found

    • lowTypeScript strict mode on

      Strict typing gives an agent typed signatures to anchor an edit on, not guesses.

      Fix: Turn on `strict` in tsconfig.json.

    • Conventional source layout β€” A recognised src/ Β· lib/ Β· app/ layout lets an agent localise code without spelunking.
  • Verification Surface

    5 checks passed2 problems found

    • mediumTest command defined

      A `test` script is the documented, discoverable way to run the suite.

      Fix: Add a `test` script (package.json / Makefile / Taskfile) that runs the suite.

    • lowTask runner present

      A Makefile / Taskfile / justfile gives an agent a single index of the project's commands.

      Fix: Add a Makefile / Taskfile / justfile collecting the common commands.

    • Tests present β€” Runnable tests are the ground truth an agent checks its own edits against.
    • Lint command defined β€” A linter an agent can run catches a class of mistakes before they ship.
    • CI configured β€” Readable CI is the feedback loop an agent reads to know what 'green' means here.
    • Typecheck command β€” A typecheck script lets an agent catch type errors before they ship.
    • Build command β€” A documented build command is the discoverable way for an agent to produce artifacts.
  • Agent Safety

    8 checks passed0 problems found

    • No committed .env β€” A tracked real .env hands secrets to anything β€” including an agent β€” that reads the repo.
    • No exposed secrets β€” Secret-shaped literals in tracked files are a live credential leak.
    • .gitignore covers env files β€” An ignore rule for .env is what keeps the next careless commit from leaking secrets.
    • CODEOWNERS present β€” CODEOWNERS marks the 'don't touch without review' zones an agent should respect.
    • Automated dependency updates β€” Dependabot / Renovate keeps dependencies patched without a human in the loop.
    • .gitignore present β€” A .gitignore is the first guard against an agent committing secrets or build junk.
    • Security policy β€” A SECURITY.md is where an autonomous security agent reads this repo's rules before acting on it.
    • License makes the code reusable β€” Without a license an agent cannot safely copy, redistribute, or vendor any of this code.
  • Context Economy

    3 checks passed1 problem found

    • lowGenerated / vendored files marked

      A .gitattributes linguist marker tells tools (and agents) what to skip.

      Fix: Mark generated/vendored paths in .gitattributes (linguist-generated / linguist-vendored).

    • Dependencies not vendored in β€” A committed node_modules / vendor tree drowns the real signal an agent needs.
    • Build output not committed β€” Generated dist/ Β· build/ output is noise an agent has to read past.
    • No oversized source file β€” A single file in the hundreds of KB blows past an agent's effective context window.

Where the grip slips

  1. highPath To Contribute+15 pts if fixed

    PR restrictions for non-contributors block the typical contribution path

    The README and CONTRIBUTING both state that PRs are restricted to contributors only, and the CONTRIBUTING guide requires issue discussion before coding most changes. For a newcomer, this means they cannot follow the standard 'find an issue β†’ code β†’ open a PR' workflow. The barrier to becoming a contributor is never explained, so a newcomer who picks up a 'good first issue' may code a fix only to discover the PR will be rejected.

    Fix: Clarify in CONTRIBUTING.md exactly what defines a 'contributor' and what the process is to earn contributor status (e.g., first PR review, maintainer approval). Consider reopening PRs from new contributors once v2 stabilizes, or create a clear alternate path (e.g., 'open a draft PR for feedback before implementation'). If restrictions must remain, escalate this to the top of README to set expectations upfront.

  2. highGetting Running+12 pts if fixed

    No explicit local monorepo development setup instructions

    The README and CONTRIBUTING guide code and run examples from published packages (npm install @modelcontextprotocol/server), but there are no instructions for cloning the repo and setting up a local development environment. A newcomer wanting to contribute must infer the setup (pnpm install, pnpm build, pnpm test) from package.json scripts without guidance. This creates friction when debugging or testing changes.

    Fix: Add a 'Development Setup' section to CONTRIBUTING.md with: (1) clone + pnpm@10.26.1 requirement, (2) pnpm install && pnpm build commands, (3) how to run tests (pnpm test:all), (4) how to build and test locally before submitting a PR. Optionally link to a DEVELOPMENT.md file with more detail.

  3. mediumFinding Your Way+10 pts if fixed

    No monorepo architecture overview or structure map

    The README lists package names and middleware, and file tree shows directories, but there is no visual or narrative map of how the monorepo is organized. A newcomer exploring the codebase to understand the project structure must infer boundaries between packages/client, packages/server, examples/, middleware/, and internal tooling. This is especially important because v2 introduces breaking changes from v1.

    Fix: Add a 'Repository Structure' section to README or a new ARCHITECTURE.md file that explains: (1) monorepo layout (pnpm workspaces), (2) each package's purpose (client, server, middleware, test-conformance, codemod), (3) how examples and docs are organized, (4) how to navigate from a problem statement to the right package/file.

  4. mediumFirst Impression+8 pts if fixed

    v1 to v2 migration path is underdocumented in the README

    The README heavily emphasizes that v1.x is the stable version and v2 is pre-alpha, and links to v1 API docs, but does not explain to a v1 user or newcomer how/when they should migrate or what has changed. The file tree shows migration.md and migration-SKILL.md, but these are buried and not linked in the README overview.

    Fix: Add a subsection in the README under 'Getting Started' or 'Documentation' that says: 'Using v1? See the [migration guide](docs/migration.md) for v2 upgrade path.' Provide a one-sentence summary of the major breaking changes (e.g., 'v2 replaces X with Y; see [migration.md](docs/migration.md) for details').

  5. mediumGetting Running+8 pts if fixed

    No lockfile visible; pnpm version is not pinned in a standard way

    The package.json specifies `"packageManager": "pnpm@10.26.1"`, which is good, but the snapshot does not show pnpm-lock.yaml or any other lockfile in the file tree. This makes it unclear whether the dependency graph is reproducible across environments, and a newcomer may clone the repo and hit different transitive versions of dependencies.

    Fix: Ensure pnpm-lock.yaml is committed to the repository and visible in the file tree. If it already exists but was excluded from the snapshot, no action is needed, but verify this is documented. If not, run `pnpm install` once and commit the lockfile.

  6. lowFirst Impression+5 pts if fixed

    Multiple pre-alpha warnings risk overwhelming newcomers

    The README opens with two large warning blocks (IMPORTANT: v2 is pre-alpha; WARNING: PRs restricted). While honesty is good, the cumulative effect may discourage newcomers from engaging. The pre-alpha status is stated, but no clear 'unstable APIs' or 'here's how to use stable v1' quick link is provided in the same breath.

    Fix: Consolidate the two warnings into one concise, actionable banner at the top: 'This is v2 (pre-alpha, stable Q3 2026). Use v1.x for production. PRs from contributors only; see [CONTRIBUTING.md](CONTRIBUTING.md). Questions? [Start here](docs/faq.md).' Link to FAQ or getting-started docs to guide newcomers next steps.

  7. lowOpen Door+3 pts if fixed

    No code of conduct linked in the README

    CODE_OF_CONDUCT.md exists in the file tree, but the README does not reference it or any community guidelines. A newcomer looking for the code of conduct has to infer its existence from the file tree.

    Fix: Add a 'Code of Conduct' section or footer link in the README pointing to CODE_OF_CONDUCT.md, e.g., 'We are committed to fostering an open and respectful community. See our [Code of Conduct](CODE_OF_CONDUCT.md).'

Show your score

Latch badge
[![Latch score](https://latch.fyi/api/badge/525cb81c-c82f-4cff-b5bf-0703ef5f3666)](https://latch.fyi/modelcontextprotocol/typescript-sdk)

Share this analysis

https://latch.fyi/modelcontextprotocol/typescript-sdk

Fixed some issues? Re-analyze to measure your progress β†’

Copy the whole report

Grab the entire verdict in one go β€” a Markdown block for a README or an issue, plain text, or a prompt you can paste straight into an LLM.

Was this verdict fair?

One tap helps β€” or tell us more.