@ianyian/myskills 0.1.1 → 0.2.1

package/README.md

@@ -1,78 +1,49 @@
 # @ianyian/myskills
 
-Three practical AI skills that turn a Node.js CLI into a small project assistant:
+Portable AI skills for VS Code Copilot Agent mode:
 
-- **`repo-explorer`** understands a local repository and writes an architecture and onboarding guide.
-- **`research-brief`** turns a topic and optional URLs into a cited Markdown research brief.
-- **`idea-to-execution`** converts an idea into an implementation plan with tasks and acceptance criteria.
+- **repo-explorer** - creates a repository architecture and onboarding guide.
+- **research-brief** - creates a concise research brief with citations.
+- **idea-to-execution** - turns an idea into an implementation plan.
 
 Repository: <https://github.com/ianyian/mySKILLs>
 
-## Requirements
+## Use in a project
 
-- Node.js 18 or newer
-- An OpenAI-compatible API key
+Requirements: Node.js 18+ and VS Code with GitHub Copilot Agent mode.
 
-The CLI uses the OpenAI-compatible Chat Completions API. It works with OpenAI by default and can be pointed at another compatible provider with `OPENAI_BASE_URL`.
-
-## Install and configure
-
-Run without installing:
-
-```bash
-npx @ianyian/myskills --help
-```
-
-Or install globally:
-
-```bash
-npm install --global @ianyian/myskills
-```
-
-Configure the model client:
-
-```bash
-export OPENAI_API_KEY="your-api-key"
-# Optional:
-export OPENAI_MODEL="gpt-4o-mini"
-export OPENAI_BASE_URL="https://api.openai.com/v1"
-```
-
-## Examples
-
-Analyze the current repository:
+From the project folder, run:
 
 ```bash
-npx @ianyian/myskills repo-explorer .
+npx @ianyian/myskills@latest init
 ```
 
-This creates `PROJECT_OVERVIEW.md`. Choose another output path with `--output`:
+This creates:
 
-```bash
-npx @ianyian/myskills repo-explorer ./my-project --output docs/onboarding.md
+```text
+.github/skills/
+├── repo-explorer/SKILL.md
+├── research-brief/SKILL.md
+└── idea-to-execution/SKILL.md
 ```
 
-Create a research brief:
+Then open Copilot Chat in Agent mode and ask:
 
-```bash
-npx @ianyian/myskills research-brief "Compare PostgreSQL and MongoDB"
+```text
+Use the repo-explorer skill to create an onboarding guide for this repository.
 ```
 
-Add source URLs. The skill fetches their text and asks the model to cite them:
+The other skills can be used with requests such as:
 
-```bash
-npx @ianyian/myskills research-brief \
-  "Best practices for Node.js error handling" \
-  --url https://nodejs.org/en/learn/getting-started/introduction-to-nodejs \
-  --output docs/node-research.md
+```text
+Use the research-brief skill to compare PostgreSQL and MongoDB, with citations.
+Use the idea-to-execution skill to plan dark mode for this web application.
 ```
 
-Turn a product idea into an execution plan:
+The agent creates the requested Markdown artifacts in your project. Use `--force` to replace existing skill files:
 
 ```bash
-npx @ianyian/myskills idea-to-execution \
-  "Add dark mode to my web application" \
-  --output implementation-plan.md
+npx @ianyian/myskills@latest init --force
 ```
 
 ## Development
@@ -83,19 +54,16 @@ npm test
 npm pack --dry-run
 ```
 
-The package contains no runtime dependencies; it uses Node.js built-ins and `fetch`.
-
-## Publish
-
-After logging in to npm, publish a new version:
+## Automated npm publishing
 
-```bash
-npm login
-npm publish --access public
-```
+Publishing runs automatically after a pull request is merged into `main`.
 
-Every later publish needs a new version, for example `npm version patch && npm publish`.
+1. In npm package settings, configure a **Trusted Publisher** for GitHub Actions with owner `ianyian`, repository `mySKILLs`, workflow filename `publish.yml`, and no environment. The filename is case-sensitive and must be the filename only, not the `.github/workflows/` path.
+2. Merge a pull request into `main`.
+3. GitHub Actions runs `npm test`.
+4. If the version already exists on npm, the workflow increments the patch version.
+5. GitHub publishes the package with OIDC provenance, then commits the new version to `main`.
 
-## Security and privacy
+The workflow uses npm 11.5.1 for Trusted Publishing, ignores its own release commit, and creates one npm release per merge. Published npm versions are immutable. Manual releases require a version bump before `npm publish`.
 
-The CLI sends the repository excerpts or fetched source text to the configured model provider. Review your provider's data policy before using it with private code. API keys are read from environment variables and are never written to generated files.
+The normal Agent Skills workflow does not require an API key. The package also retains an optional standalone CLI mode for OpenAI-compatible providers, which requires `OPENAI_API_KEY`.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ianyian/myskills",
-  "version": "0.1.1",
+  "version": "0.2.1",
   "description": "Practical AI skills for repository analysis, research, and execution planning",
   "type": "module",
   "bin": {
@@ -8,6 +8,7 @@
   },
   "files": [
     "bin",
+    "skills",
     "src",
     "README.md",
     "LICENSE"

package/skills/idea-to-execution/SKILL.md

@@ -0,0 +1,25 @@
+---
+name: idea-to-execution
+description: Turn a product or engineering idea into an implementation-ready plan.
+---
+
+# Idea to Execution
+
+Use this skill when the user has a rough idea and wants an actionable engineering plan.
+
+## Workflow
+
+1. Restate the problem and intended outcome.
+2. Make assumptions explicit and identify non-goals.
+3. Inspect the repository when the idea relates to an existing codebase.
+4. Produce `implementation-plan.md` unless the user requests another output path.
+5. Include:
+   - Problem and goals
+   - User experience
+   - Proposed technical approach
+   - Ordered tasks with likely files or modules
+   - Acceptance criteria
+   - Testing strategy
+   - Risks and open questions
+
+Keep tasks small enough to implement and verify independently. Do not silently make product decisions when multiple reasonable options exist.

package/skills/repo-explorer/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: repo-explorer
+description: Understand a repository and create a practical architecture and onboarding guide.
+---
+
+# Repository Explorer
+
+Use this skill when the user asks you to understand, document, or onboard someone to a repository.
+
+## Workflow
+
+1. Inspect the repository structure, package manifests, configuration, entry points, tests, and documentation.
+2. Trace the main execution or data flow. Prefer evidence from source code over assumptions.
+3. Identify the commands needed to install, run, test, and build the project.
+4. Write `PROJECT_OVERVIEW.md` unless the user requests another output path.
+5. Include:
+   - Executive summary
+   - Architecture and data flow
+   - Important directories and files
+   - Setup and development commands
+   - Dependencies and configuration
+   - Risks, unknowns, and recommended next steps
+
+Do not invent behavior or claim that a file exists without inspecting it. Keep the guide concise enough to be useful during onboarding.

package/skills/research-brief/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: research-brief
+description: Research a topic using available web tools and produce a concise cited Markdown brief.
+---
+
+# Research Brief
+
+Use this skill when the user asks to compare technologies, investigate a topic, or summarize external information.
+
+## Workflow
+
+1. Clarify the scope and the decision the research should support.
+2. Prefer primary and authoritative sources. Use the user's URLs when supplied.
+3. Separate verified facts, expert recommendations, and interpretation.
+4. Produce `research-brief.md` unless the user requests another output path.
+5. Include:
+   - Executive summary
+   - Key findings
+   - Comparison and trade-offs where relevant
+   - Practical recommendation
+   - Sources with inline citations
+   - Open questions or claims requiring verification
+
+Do not imply that a source was consulted if it was not. Include access dates when the agent's available tooling supports them.

package/src/cli.js

@@ -1,6 +1,7 @@
 import { runRepoExplorer } from "./skills/repo-explorer.js";
 import { runResearchBrief } from "./skills/research-brief.js";
 import { runIdeaToExecution } from "./skills/idea-to-execution.js";
+import { installSkills } from "./install-skills.js";
 
 const HELP = `@ianyian/myskills - practical AI skills for projects
 
@@ -11,6 +12,7 @@ Commands:
   repo-explorer [directory]       Analyze a repository
   research-brief <topic>         Create a research brief
   idea-to-execution <idea>        Create an implementation plan
+  init [--force]                  Install VS Code Agent Skills in this project
 
 Options:
   --output <file>                 Output Markdown path
@@ -32,5 +34,6 @@ export async function run(args) {
   if (command === "repo-explorer") return runRepoExplorer(rest);
   if (command === "research-brief") return runResearchBrief(rest);
   if (command === "idea-to-execution") return runIdeaToExecution(rest);
+  if (command === "init") return installSkills({ force: rest.includes("--force") });
   throw new Error(`Unknown command "${command}". Run "myskills --help" for usage.`);
 }

package/src/install-skills.js

@@ -0,0 +1,29 @@
+import { access, cp, mkdir } from "node:fs/promises";
+import path from "node:path";
+import { fileURLToPath } from "node:url";
+
+const packageRoot = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..");
+const sourceRoot = path.join(packageRoot, "skills");
+
+const skillNames = ["repo-explorer", "research-brief", "idea-to-execution"];
+
+export async function installSkills({ force = false } = {}) {
+  const destinationRoot = path.resolve(".github", "skills");
+  await mkdir(destinationRoot, { recursive: true });
+
+  for (const skillName of skillNames) {
+    const source = path.join(sourceRoot, skillName);
+    const destination = path.join(destinationRoot, skillName);
+    if (!force) {
+      try {
+        await access(destination);
+        throw new Error(`${destination} already exists. Use --force to replace it.`);
+      } catch (error) {
+        if (error.code !== "ENOENT") throw error;
+      }
+    }
+    await cp(source, destination, { recursive: true, force });
+  }
+
+  console.log(`Installed ${skillNames.length} VS Code Agent Skills in ${destinationRoot}`);
+}