#601

Global Rank · of 601 Skills

npm-publish AI Agent Skill

View Source: b-open-io/prompts

Safe

Installation

npx skills add b-open-io/prompts --skill npm-publish

84

Installs

npm-publish

MANDATORY — Read Before Doing Anything

NEVER ask the user for an OTP code. Auth is handled by scripts + agent-browser.

NEVER run manual npm/bun commands like npm whoami, npm view, bun publish, or npm publish.

You MUST run these scripts. Do NOT skip steps.

Step 1: Preflight

bash ${SKILL_DIR}/scripts/preflight.sh

Handles deterministically: version check against npm registry, bump if needed (resets gaps), build, commit log output. Pass minor or major to override default patch bump.

Step 2: Write Changelog

Read the commit log from preflight output. If CHANGELOG.md exists, add entry at top matching existing format. If not, create one. Use the version from preflight output. Categorize: Breaking Changes, Added, Changed, Fixed, Security, Deprecated.

Step 3: Release (commit + push + publish)

bash ${SKILL_DIR}/scripts/release.sh [--access public]

Commits, pushes, then calls publish.sh. If publish.sh outputs PUBLISH_SUCCESS — done, go to Step 4.

If publish.sh outputs AUTH_FAILED

The agent must orchestrate token setup. Do NOT call setup-token.sh as one long command. Run it in two phases with user communication between them.

Phase 1 — Fill the form:

bash ${SKILL_DIR}/scripts/setup-token.sh fill

Status codes:

  • FORM_READY:<username> — form is filled in Chrome, proceed to tell user
  • NOT_LOGGED_IN — tell user: "Sign in to npmjs.com in Chrome, then I'll retry"
  • FORM_NOT_FOUND — tell user: "Could not find the token form. The page may have changed."

After getting FORM_READY, tell the user directly (not inside a bash command):

I've opened the npm token creation form in Chrome and filled it out (cli-publish, 7-day, read+write, all packages). Scroll down and click Generate token when ready.

Phase 2 — Capture the token:

bash ${SKILL_DIR}/scripts/setup-token.sh capture

This polls until the token appears on the page, clicks the Copy button, reads from clipboard, writes to ~/.npmrc, and clears clipboard. The token never appears in terminal output.

Status codes:

  • TOKEN_SAVED — success, retry publish
  • CAPTURE_TIMEOUT — tell user: "Could not capture token. Copy it from Chrome and I'll write it to ~/.npmrc"

After TOKEN_SAVED, retry publish:

bash ${SKILL_DIR}/scripts/publish.sh [--access public]

Tell user: "Complete the OTP checkbox in your browser if prompted."

Step 4: Verify (background)

bash ${SKILL_DIR}/scripts/verify.sh <package-name> <version>

Run with run_in_background: true. Exponential backoff (5s, 10s, 20s, 40s, 60s).

Key Architecture Principle

Scripts output status codes. The agent interprets them and talks to the user. Script output is hidden inside collapsed bash commands — the user won't see it. All user-facing communication must be direct agent messages OUTSIDE of bash calls.

Installs

Installs 84
Global Rank #601 of 601

Security Audit

ath Safe
socket Safe
Alerts: 0 Score: 90
snyk Low
EU EU-Hosted Inference API

Power your AI Agents with the best open-source models.

Drop-in OpenAI-compatible API. No data leaves Europe.

Explore Inference API

GLM

GLM 5

$1.00 / $3.20

per M tokens

Kimi

Kimi K2.5

$0.60 / $2.80

per M tokens

MiniMax

MiniMax M2.5

$0.30 / $1.20

per M tokens

Qwen

Qwen3.5 122B

$0.40 / $3.00

per M tokens

How to use this skill

1

Install npm-publish by running npx skills add b-open-io/prompts --skill npm-publish in your project directory. Run the install command above in your project directory. The skill file will be downloaded from GitHub and placed in your project.

2

No configuration needed. Your AI agent (Claude Code, Cursor, Windsurf, etc.) automatically detects installed skills and uses them as context when generating code.

3

The skill enhances your agent's understanding of npm-publish, helping it follow established patterns, avoid common mistakes, and produce production-ready output.

What you get

Skills are plain-text instruction files — not executable code. They encode expert knowledge about frameworks, languages, or tools that your AI agent reads to improve its output. This means zero runtime overhead, no dependency conflicts, and full transparency: you can read and review every instruction before installing.

Compatibility

This skill works with any AI coding agent that supports the skills.sh format, including Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider, and other tools that read project-level context files. Skills are framework-agnostic at the transport level — the content inside determines which language or framework it applies to.

Data sourced from the skills.sh registry and GitHub. Install counts and security audits are updated regularly.

EU Made in Europe

Chat with 100+ AI Models in one App.

Use Claude, ChatGPT, Gemini alongside with EU-Hosted Models like Deepseek, GLM-5, Kimi K2.5 and many more.

Get the App:

Customer Support