AgentSkillsCN

bun

为脚本、测试与软件包工作流制定Bun运行时与工具使用规范。

SKILL.md
--- frontmatter
name: bun
description: Bun runtime and tooling conventions for scripts, tests, and package workflows.
version: 1.0.0
stability: stable
tier: pack
tags: [bun, runtime, test, scripts]
anti_tags: [npm-only]
tools:
  required: [bash]
  optional: [read, lsp_diagnostics, skill_complete]
  denied: []
budget:
  max_tool_calls: 35
  max_tokens: 80000
outputs: [bun_commands_run, environment_notes]
consumes: []
escalation_path: {}

Bun Pack Skill

Intent

Use Bun-native workflows consistently for install, build, test, and script execution.

Trigger

Use this pack when project runtime or package tooling is Bun-based.

Core Rules

  • •Prefer bun commands over npm/yarn equivalents.
  • •Keep script invocations deterministic and workspace-aware.
  • •Match repository script names from package.json instead of inventing commands.

Command Conventions

Install and dependency operations

bash
bun install

Run package scripts

bash
bun run <script-name>

Run tests

bash
bun test
bun test <target>

Execute TypeScript/JavaScript files

bash
bun run <file.ts>

Workflow

Step 1: Discover available scripts

Inspect root and relevant package manifests before running commands.

Step 2: Pick smallest useful command

  • •local verification: targeted script/test first
  • •broader confidence: full script/test suite second

Step 3: Record environment notes

Report environment caveats when commands depend on:

  • •missing .env variables
  • •unavailable external services
  • •workspace/package filters

Stop Conditions

  • •Repository is not Bun-based and no Bun workflow is configured.
  • •Required script is absent and there is no documented equivalent.
  • •Command execution depends on unavailable environment or service.

Anti-Patterns (never)

  • •Using npm commands in Bun-first repos without explicit reason.
  • •Running heavyweight full suites before targeted validation.
  • •Assuming script names (test, build) without checking manifests.
  • •Hiding command failures behind generic summaries.

Example

Input:

text
"Validate TypeScript changes in this Bun monorepo."

Expected sequence:

  1. •inspect script names in package manifests.
  2. •run bun run typecheck (or repo-specific equivalent).
  3. •run targeted and then broader tests as needed.
  4. •report bun_commands_run with evidence and caveats.