update-skills
Create or update repository skills and instructions when major learnings are discovered during a session. Use when the user says "learn!", when a significant pattern or pitfall is identified, or when reusable domain knowledge should be captured for future sessions.
- License 许可
- MIT
- 最近更新
- 2026-11-10
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- update-skills
- description
- Create or update repository skills and instructions when major learnings are discovered during a session. Use when the user says "learn!", when a significant pattern or pitfall is identified, or when reusable domain knowledge should be captured for future sessions.
03 SKILL.md 正文
<!-- Customize this skill and select save to override its behavior. Delete that copy to restore the built-in behavior. -->
Update Skills & Instructions
When a major repository learning is discovered — a recurring pattern, a non-obvious pitfall, a crucial architectural constraint, or domain knowledge that would save future sessions significant time — capture it as a skill or instruction so it persists across sessions.
When to Use
- The user explicitly says "learn!" or asks to capture a learning
- You discover a significant pattern or constraint that cost meaningful debugging time
- You identify reusable domain knowledge that isn't documented anywhere in the repo
- A correction from the user reveals a general principle worth preserving
Decision: Skill vs Instruction vs Learning
Add a learning to an existing instruction when:
- The insight is small (1-4 sentences) and fits naturally into an existing instruction file
- It refines or extends an existing guideline
- Follow the pattern in
.github/instructions/learnings.instructions.md
Create or update a skill (.github/skills/{name}/SKILL.md or .agents/skills/{name}/SKILL.md) when:
- The knowledge is substantial (multi-step procedure, detailed guidelines, or rich examples)
- It covers a distinct domain area (e.g., "how to debug X", "patterns for Y")
- Future sessions should be able to invoke it by name
Create or update an instruction (.github/instructions/{name}.instructions.md) when:
- The rule should apply automatically based on file patterns (
applyTo) or globally - It's a coding convention, architectural constraint, or process rule
- It doesn't need to be invoked on demand
Procedure
1. Identify the Learning
Reflect on what went wrong or what was discovered:
- What was the problem or unexpected behavior?
- Why was it a problem? (root cause, not symptoms)
- How was it fixed or what's the correct approach?
- Can it be generalized beyond this specific instance?
2. Check for Existing Files
Before creating new files, search for existing skills and instructions that might be the right home:
# Check existing skills
ls .github/skills/ .agents/skills/ 2>/dev/null
# Check existing instructions
ls .github/instructions/ 2>/dev/null
# Search for related content
grep -r "related-keyword" .github/skills/ .github/instructions/ .agents/skills/
3a. Add to Existing File
If an appropriate file exists, add the learning to its ## Learnings section (create the section if it doesn't exist). Each learning should be 1-4 sentences.
3b. Create a New Skill
If the knowledge warrants a standalone skill:
- Choose the location:
.github/skills/{name}/SKILL.mdfor project-level skills (committed to repo).agents/skills/{name}/SKILL.mdfor agent-specific skills
- Create the directory and SKILL.md with frontmatter:
---
name: {skill-name}
description: {One-line description of when and why to use this skill.}
---
# {Skill Title}
{Body with guidelines, procedures, examples, and learnings.}
- The
namefield must match the parent folder name exactly. - Include concrete examples — skills with examples are far more useful than abstract rules.
3c. Create a New Instruction
If the knowledge should apply automatically:
---
description: {When these instructions should be loaded}
applyTo: '{glob pattern}' # optional — auto-load when matching files are attached
---
{Content of the instruction.}
4. Quality Checks
Before saving:
- Is the learning general enough to help future sessions, not just this one?
- Is it specific enough to be actionable, not just a vague principle?
- Does it include a concrete example of right vs wrong?
- Does it avoid duplicating knowledge already captured elsewhere?
- Is the description clear enough that the agent will know when to invoke/apply it?
5. Inform the User
After creating or updating the file:
- Summarize what was captured and where
- Explain why this location was chosen
- Note if any existing content was updated vs new content created
04 同仓库更多 Skills
vercel-sandbox
Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include "Vercel Sandbox browser", "microVM Chrome", "agent-browser in sandbox", "browser automation on Vercel", or any task requiring Chrome in a Vercel Sandbox.
slack
Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include "check my Slack", "what channels have unreads", "send a message to", "search Slack for", "extract from Slack", "find who said", or any task requiring programmatic Slack interaction.
electron
Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include "automate Slack app", "control VS Code", "interact with Discord app", "test this Electron app", "connect to desktop app", or any task requiring automation of a native Electron application.
dogfood
Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to "dogfood", "QA", "exploratory test", "find issues", "bug hunt", "test this app/site/platform", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.
agent-browser
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
shadcn
Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn/ui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for "shadcn init", "create an app with --preset", "switch to --preset", and for building a design system or applying a DESIGN.md or brand spec to shadcn/ui.