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

A professionally maintained SDK with clear purpose and solid docs, but setup friction and contributing barriers cost it points.

TypeScript★ 1,999113 open issuesMIT License
Overall score: 78 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.

26 of 35 checks passed across 6 categories.

  • Machine Orientation

    5 checks passed3 problems found

    • lowllms.txt present

      An llms.txt is the emerging machine-orientation standard — 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 — An agent reconstructs what the repo is and how to work in it from the README first.
    • Agent instructions file — AGENTS.md / CLAUDE.md is the direct, machine-read channel for entry points and conventions.
    • Agent instructions stay lean — A bloated instructions file burns context and measurably hurts more than it helps.
    • Repository description set — The one-line description is the fastest orientation signal an agent reads.
    • Contribution conventions are written down — A contributor guide is where an agent learns the branch, commit, and review conventions it must follow.
  • Reproducible Setup

    5 checks passed1 problem found

    • 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.

    • Lockfile committed — Without a lockfile two clones can resolve different dependency trees — the build isn't reproducible.
    • Runtime / toolchain pinned — An unpinned runtime makes an agent guess the version a clean environment needs.
    • Version pins are exact — A floating pin (a bare major, or a range) still drifts between installs.
    • Container or dev-container — A Dockerfile / compose / devcontainer removes 'works on my machine' from setup.
    • Dev container — A .devcontainer gives an agent a one-command, fully-declared environment.
  • Code Navigability

    2 checks passed0 problems found

    • Conventional source layout — A recognised src/ · lib/ · app/ layout lets an agent localise code without spelunking.
    • TypeScript strict mode on — Strict typing gives an agent typed signatures to anchor an edit on, not guesses.
  • Verification Surface

    5 checks passed2 problems found

    • 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.

    • lowTypecheck command

      A typecheck script lets an agent catch type errors before they ship.

      Fix: Add a `typecheck` script (e.g. `tsc --noEmit`).

    • Tests present — Runnable tests are the ground truth an agent checks its own edits against.
    • Test command defined — A `test` script is the documented, discoverable way to run the suite.
    • Lint command defined — A linter an agent can run catches a class of mistakes before they ship.
    • CI configured — Readable CI is the feedback loop an agent reads to know what 'green' means here.
    • Build command — A documented build command is the discoverable way for an agent to produce artifacts.
  • Agent Safety

    6 checks passed2 problems found

    • medium.gitignore covers env files

      An ignore rule for .env is what keeps the next careless commit from leaking secrets.

      Fix: Add `.env` (and `.env.*`) to .gitignore.

    • lowAutomated dependency updates

      Dependabot / Renovate keeps dependencies patched without a human in the loop.

      Fix: Enable Dependabot or Renovate.

    • No committed .env — A tracked real .env hands secrets to anything — including an agent — that reads the repo.
    • No exposed secrets — Secret-shaped literals in tracked files are a live credential leak.
    • CODEOWNERS present — CODEOWNERS marks the 'don't touch without review' zones an agent should respect.
    • .gitignore present — A .gitignore is the first guard against an agent committing secrets or build junk.
    • Security policy — A SECURITY.md is where an autonomous security agent reads this repo's rules before acting on it.
    • License makes the code reusable — 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 — A committed node_modules / vendor tree drowns the real signal an agent needs.
    • Build output not committed — Generated dist/ · build/ output is noise an agent has to read past.
    • No oversized source file — A single file in the hundreds of KB blows past an agent's effective context window.

Where the grip slips

  1. highGetting Running+12 pts if fixed

    Mock server setup for testing is undocumented and requires external tool

    CONTRIBUTING.md tells developers to run `./scripts/mock` and then `yarn test`, citing a requirement for a mock server against OpenAPI spec using an external tool called 'steady'. However, there is no install guide for 'steady', no CI config showing how the CI actually runs tests, and no fallback or guidance for running tests without the mock server. A newcomer will run `yarn test` and hit an error with no path forward.

    Fix: Document the mock server setup in CONTRIBUTING.md: add a 'Mock Server Setup' section with links to install 'steady', instructions to run `./scripts/mock`, and explain what happens if tests fail due to mock server issues. Alternatively, show how CI runs tests in the `ci.yml` workflow and make that the source of truth.

  2. highFinding Your Way+11 pts if fixed

    No explanation of four sub-packages and when to use each

    The file tree shows four separate SDKs in `packages/` (aws-sdk, bedrock-sdk, foundry-sdk, vertex-sdk), each with their own package.json and CI config. The README and CONTRIBUTING.md do not mention these exist, their purpose, or when a user should reach for one vs. the main package. A newcomer exploring the repo will be confused by the structure and may not realize these are separate entry points for different deployment targets.

    Fix: Add a 'Monorepo Structure' section to the README or CONTRIBUTING.md explaining the four sub-packages, their purpose (e.g., foundry-sdk for Anthropic Foundry, aws-sdk for AWS Bedrock), and when a contributor should modify which one. Link each to its own README if those exist and are helpful.

  3. mediumFinding Your Way+8 pts if fixed

    Generated vs. hand-written code is not labeled or distinguishable

    CONTRIBUTING.md states that 'most of the SDK is generated code' and that modifications 'may result in merge conflicts' and the generator 'will never modify the contents of `src/lib/` and `examples/` directories.' However, a newcomer reading the source tree has no way to tell which files are generated and which are safe to edit. There is no comment header, no .generated-marker, and no listing of which directories are generated.

    Fix: Add a comment block at the top of all generated files (e.g., `src/api.ts`, `src/api-promise.ts`, generated client files) stating 'This file is generated and should not be manually edited' with a link to CONTRIBUTING.md. Alternatively, create a `GENERATED_FILES.md` listing all generated paths and an explanation of the generation process.

  4. mediumPath To Contribute+7 pts if fixed

    No PR or issue templates to guide contributors

    The snapshot does not show GitHub issue or pull request templates (normally `.github/ISSUE_TEMPLATE/` and `.github/PULL_REQUEST_TEMPLATE.md`). Without these, contributors may submit incomplete PRs or issues, and maintainers have no structured way to prompt for required info (e.g., SDK version, reproduction steps, which example doesn't work).

    Fix: Create `.github/ISSUE_TEMPLATE/bug_report.yml` and `.github/ISSUE_TEMPLATE/feature_request.yml` with fields for SDK version, environment, minimal repro, and expected vs. actual behavior. Create `.github/pull_request_template.md` with a checklist for tests, docs updates, and changelog entries.

  5. mediumGetting Running+6 pts if fixed

    Build from source has hidden gotchas in the prepare hook

    The `package.json` includes a `prepare` script that runs shell scripts checking git state and conditionally building/swapping symlinks. This is not documented in CONTRIBUTING.md and will cause confusion when a developer clones the repo, runs `yarn`, and sees unexpected behavior. The `check-is-in-git-install.sh` is cryptic—a developer with no context may think the build failed.

    Fix: Add a note to CONTRIBUTING.md under 'Setting up the environment' explaining that `yarn` will run a prepare hook that checks git state; clarify what the hook does and why (e.g., 'to ensure dist/ stays in sync'). If the prepare hook is only for CI/CD, move it out of package.json and into a separate CI step.

  6. lowPath To Contribute+5 pts if fixed

    No beginner-friendly issues or contribution roadmap

    The repo has 113 open issues but the snapshot does not show any labels like 'good-first-issue', 'help-wanted', or 'beginner-friendly'. New contributors have no signal of which issues are safe or suitable for first contributions. There is also no contribution roadmap or list of high-value contributions Anthropic would like.

    Fix: Use GitHub issue labels to mark 5–10 beginner-friendly issues as 'good-first-issue' or 'help-wanted'. Add a 'Contributing' section to the README or a `CONTRIBUTING.md` header noting that newcomers should look for these labels. Optionally, maintain a simple roadmap in the repo documenting high-priority gaps or features.

  7. lowPath To Contribute+5 pts if fixed

    Examples are not easily executable without knowing the shebang pattern

    The CONTRIBUTING.md shows that examples like `examples/<your-example>.ts` require a shebang (`#!/usr/bin/env -S npm run tsn -T`) and `chmod +x` to run as executable scripts. This is clever but not intuitive—a newcomer may not know to make the file executable or may try to run `node examples/demo.ts` and get an error. Only the shebang pattern is shown; there's no explanation of why this is necessary or when to use it.

    Fix: Add an 'Examples' section to CONTRIBUTING.md or README.md explaining: (1) examples have a shebang and require `chmod +x` to run as executables, (2) the pattern `#!/usr/bin/env -S npm run tsn -T` uses ts-node to run TypeScript directly, (3) how to run an example (`./examples/demo.ts` vs. `yarn tsn examples/demo.ts` vs. `node dist/examples/demo.js`). Provide a copy-paste template for new examples.

  8. lowOpen Door+4 pts if fixed

    No code of conduct or community guidelines

    The file tree contains SECURITY.md and CONTRIBUTING.md but no CODE_OF_CONDUCT.md. This signals to contributors that the project may not have clear norms for respectful participation, which is especially important for a widely-used SDK.

    Fix: Add a CODE_OF_CONDUCT.md (e.g., based on Contributor Covenant) at the repo root and link it from CONTRIBUTING.md and README.md.

Show your score

Latch badge
[![Latch score](https://latch.fyi/api/badge/a764292a-c73a-4e63-804d-abf1eadc6de4)](https://latch.fyi/anthropics/anthropic-sdk-typescript)

Share this analysis

https://latch.fyi/anthropics/anthropic-sdk-typescript

Fixed some issues? Re-analyze to measure your progress →

Copy the whole report

Grab the entire verdict in one go — 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 — or tell us more.