JavaScript100%

Commit activity

Last 12 weeks · 10 commits

Full graph →

Community health

5 of 6 standards met

Community profile →
85
✓README✓License✓Contributing✓Code of Conduct○Issue Template✓PR Template

Recent PRs & issues

Active · 4 discussions · Last activity 1mo ago
See all on GitHub →
karlhorky

Description Since pnpm 11, uses pnpm's native publishing flow and handles authentication itself: https://pnpm.io/blog/releases/11.0#native-publish-flow-no-more-npm-cli-fallback With pnpm , publishing through fails with because pnpm detects a non-interactive terminal. Steps to reproduce 1. In an existing package that you can publish to npm, use pnpm 11.25.0 via : 2. Confirm the pnpm version: 3. Run and observe the publishing failure with 💥 Expected behavior works with pnpm v11+ Environment np - 12.0.1 pnpm - 11.25.0 npm - 11.6.2 Git - 2.50.1 OS - macOS Tahoe 26.6.2 (25G83)

karlhorky · 1mo ago
fregante

Description npm now asks to complete 2FA in the browser: and I find this to be better than what np does: Is the feature request related to a problem?** Compare the two workflows: 1. Press ENTER 2. Click "Use security key" (on a pre-logged-in npmjs.com) 3. Touch ID with np's: 1. Pick up phone 2. Unlock it 3. Find Authenticator 4. Locate npm on the list 5. Tap it 6. cmd-v on the computer (via Continuity, if it works, within 4 seconds) Possible implementation I don't know, but it could skip the "Press enter to continue" step and just use to open the browser and skip one more step. Bonus points: this would also resolve another issue because "the notification" is the whole browser getting focus. https://github.com/sindresorhus/np/issues/359 Issues Potentially some people don't like this new workflow

fregante · 2mo ago
cossssmin

Description Running or installing then running , I am getting this error since : Stack trace: Environment np - 12.1.0 Node.js - 24.20.0 npm - 11.19.0 Git - 2.55.0 OS - Omarchy 4.0.2-1, Linux 7.1.9-arch1-2

cossssmin · 1mo ago
carfeii

Summary , before publishing a package, prints a summary of the commits since the last release as part of its normal, interactive confirmation flow. Each commit's message is taken directly from and written to the terminal () with no filtering of control characters. Unlike a package's field (validated by 's strict naming rules) or a git branch/tag name (restricted by git's own ref-naming rules), a commit message has no character restrictions. Since running in a repository is exactly what a maintainer preparing a release does, any commit in the repository's history containing an embedded escape sequence in its message is written to the terminal and interpreted live during a routine release. POC (available upon request) Impact I validated this by creating a commit with a message containing a raw OSC title-set escape sequence, then running 's CLI entry point in that repository inside a tmux session, answering the initial confirmation prompt, and capturing the literal byte stream written to the pty via (which taps the raw bytes before tmux's own screen-buffer interpretation) up to the point the commit history renders, before any version selection or actual publish step, and inspecting it as a hex dump. The captured stream contained the exact injected sequence, unmodified, embedded in the commit message text, confirming it reaches the terminal without any stripping. I used a benign title-set marker specifically to confirm the sequence reaches output unstripped, rather than a more aggressive payload; the specific range of what an injected OSC/CSI sequence can do depends on the victim's terminal emulator and was not enumerated further. Suggested Fix Strip control characters, particularly the ESC byte, from a commit message before it's written to the terminal, the same way already rejects them in the package name field. I'd suggest filtering once in the commit-history builder in , since that's the point where it's rendered. I'm not including the exact reproduction steps in this issue, to give time to review and land a fix before the mechanics are public. The PoC is available on request. For your review and coordinated disclosure, Geo | GitHub.com/geo-chen

carfeii · 2mo ago
etlovett

Description advertises support for several package managers, including pnpm. In our codebase we prefer to use pnpm, and would like to use it exclusively so that tooling is consistent across developer's machines. However, appears unable to run successfully with 's configured to only allow pnpm. Steps to reproduce 1. Have this in your : 2. Run Expected behavior I would expect the release to be able to run successfully. Instead, fails with while trying to execute internally. Environment np - v11.2.1, which is not the latest, though I don't see a fix for this in the release notes for v11.2.2 or v12.0.0, and Claude claims to have verified that the v12.0.0 codebase still contains this bug Node.js - 24.18.0 npm - n/a, that's the point pnpm - 11.15.1 Git - 2.54.0 OS - MacOS 26.2

etlovett · 2mo ago
Structured data for AI agents

Repository: sindresorhus/np. Description: A better `npm publish` Stars: 7711, Forks: 311. Primary language: JavaScript. Languages: JavaScript (100%). License: MIT. Topics: cli, cli-app, javascript, nodejs, npm, npm-package, npm-publish, publish, yarn. Latest release: v12.1.1 (1mo ago). Open PRs: 0, open issues: 4. Last activity: 1mo ago. Community health: 85%. Top contributors: sindresorhus, itaisteinherz, SamVerschueren, tommy-mitchell, bunysae, mifi, dopecodez, sonicdoe, oligot, LitoMore and others.