docs: add agentskills.io spec fields, plugin settings, and CLAUDE.md audit checks

- skill-builder: add license, compatibility, metadata, allowed-tools fields
  and scripts/references/assets directory structure from agentskills.io spec
- compatibility-checker: add terminal sandbox, MCP discovery, autopilot,
  and plugin settings (chat.plugins.enabled/marketplaces/paths)
- setup-audit: add CLAUDE.md and .claude/agents/ to audit checklists,
  add plugins/autopilot/MCP discovery to recommended settings
- new-agent prompt: add Claude agent format (.claude/agents/) section
- new-skill prompt: add license, compatibility, metadata optional fields

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: trsdn

.github/prompts/copilot-new-agent.prompt.md

@@ -62,6 +62,14 @@ When done, list the created file path and how to select the agent in the VS Code
 
 > **Deprecated:** `infer` is deprecated. Use `user-invocable` and `disable-model-invocation` instead.
 
+## Claude agent format (alternative)
+
+VS Code also supports the Claude agent format in `.claude/agents/`:
+- Uses plain `.md` files (not `.agent.md`)
+- `tools` is a comma-separated string (e.g., `"Read, Grep, Glob, Bash"`) instead of YAML array
+- Supports `disallowedTools` as comma-separated string
+- VS Code maps Claude tool names to corresponding VS Code tools
+
 ## Reference docs
 
 - Custom agents (VS Code): https://code.visualstudio.com/docs/copilot/customization/custom-agents

.github/prompts/copilot-new-skill.prompt.md

@@ -27,6 +27,9 @@ Create a new Agent Skill in this repository.
    - Optionally set `argument-hint` for hint text when invoked as a slash command
    - Optionally set `user-invocable: false` to hide from `/` menu (still auto-loaded by Copilot)
    - Optionally set `disable-model-invocation: true` to require manual invocation only
+   - Optionally set `license` for the skill's license
+   - Optionally set `compatibility` for environment requirements (max 500 chars)
+   - Optionally set `metadata` with `author`, `version`, or other key-value pairs
 4. In the Markdown body, include:
    - What the skill accomplishes
    - When to use it (specific triggers)

.github/skills/copilot-compatibility-checker/SKILL.md

@@ -115,8 +115,17 @@ When diagnosing issues, consult these sources for the latest information:
 
 | Feature | Setting | Since Version | Status |
 |---------|---------|---------------|--------|
-| Agent plugins | N/A | Preview | **Preview** |
-| Plugin marketplaces | N/A | Preview | **Preview** |
+| Agent plugins | `chat.plugins.enabled` | Preview | **Preview** |
+| Plugin marketplaces | `chat.plugins.marketplaces` | Preview | **Preview** |
+| Local plugins | `chat.plugins.paths` | Preview | **Preview** |
+
+### CLAUDE.md / Claude Compatibility
+
+| Feature | Setting | Since Version | Status |
+|---------|---------|---------------|--------|
+| `CLAUDE.md` instructions | `chat.useClaudeMdFile` | Stable | Stable |
+| `.claude/agents/` format | Auto-discovered | Stable | Stable |
+| `.claude/rules/` instructions | `chat.instructionsFilesLocations` | Stable | Stable |
 
 ### Prompt Files
 
@@ -130,8 +139,12 @@ When diagnosing issues, consult these sources for the latest information:
 | Feature | Setting | Since Version | Status |
 |---------|---------|---------------|--------|
 | Terminal tool auto-approval | `chat.tools.autoApprove` | 1.103 | Stable |
+| Terminal sandboxing | `chat.tools.terminal.sandbox.enabled` | Preview | **Preview** |
 | MCP servers | `chat.mcp.enabled` | 1.102 | Stable |
+| MCP sandbox | `sandboxEnabled` in `mcp.json` | Preview | **Preview** |
+| MCP auto-discovery | `chat.mcp.discovery.enabled` | Preview | **Preview** |
 | Background agents | `chat.agent.background.enabled` | 1.107 | **Experimental** |
+| Autopilot mode | `chat.autopilot.enabled` | Preview | **Preview** |
 
 ## Diagnostic Steps
 
@@ -226,6 +239,9 @@ Enable all experimental features:
   "chat.mcp.enabled": true,
   "chat.useCustomAgentHooks": true,
   "chat.useClaudeMdFile": true,
+  "chat.plugins.enabled": true,
+  "chat.autopilot.enabled": true,
+  "chat.mcp.discovery.enabled": true,
   "github.copilot.chat.customAgents.showOrganizationAndEnterpriseAgents": true
 }
 ```

.github/skills/copilot-setup-audit/SKILL.md

@@ -59,6 +59,7 @@ Check for correct directory structure:
 | `.github/copilot-instructions.md` | Workspace-wide coding guidelines | **High** |
 | `.github/instructions/*.instructions.md` | File-type specific rules | Medium |
 | `AGENTS.md` (root) | Multi-agent workspace instructions | Optional |
+| `CLAUDE.md` (root, `.claude/`, `~/.claude/`) | Cross-tool compatibility | Optional |
 
 #### Instruction Quality Checks
 
@@ -84,8 +85,8 @@ Check for correct directory structure:
 
 #### Agent File Checks
 
-- [ ] All agents have `.agent.md` extension
-- [ ] Agents are in `.github/agents/` directory
+- [ ] All agents have `.agent.md` extension (or `.md` in `.claude/agents/`)
+- [ ] Agents are in `.github/agents/` directory (or `.claude/agents/` for Claude format)
 - [ ] YAML frontmatter is valid
 - [ ] `description` is present and descriptive
 - [ ] `name` is set (recommended)
@@ -217,7 +218,10 @@ Check for recommended settings:
   "chat.useNestedAgentsMdFiles": true,
   "chat.customAgentInSubagent.enabled": true,
   "chat.useCustomAgentHooks": true,
-  "chat.useClaudeMdFile": true
+  "chat.useClaudeMdFile": true,
+  "chat.plugins.enabled": true,
+  "chat.autopilot.enabled": true,
+  "chat.mcp.discovery.enabled": true
 }
 ```
 

.github/skills/copilot-skill-builder/SKILL.md

@@ -51,6 +51,10 @@ Detailed instructions, guidelines, and examples...
 | `argument-hint` | No | Hint text shown in chat input when invoked as a slash command |
 | `user-invocable` | No | Controls `/` slash command visibility (default: `true`). Set to `false` for background skills |
 | `disable-model-invocation` | No | Controls auto-loading by Copilot (default: `false`). Set to `true` for manual-only skills |
+| `license` | No | License name or reference to a bundled license file (from [agentskills.io](https://agentskills.io/) spec) |
+| `compatibility` | No | Environment requirements — intended product, system packages, network access. Max 500 chars |
+| `metadata` | No | Arbitrary key-value mapping for additional metadata (e.g., `author`, `version`) |
+| `allowed-tools` | No | Space-delimited list of pre-approved tools the skill may use (experimental) |
 
 ### Body content
 
@@ -126,14 +130,36 @@ The description determines when Copilot loads your skill. Be specific:
 ```
 .github/skills/
 └── webapp-testing/
-    ├── SKILL.md              # Main skill file
-    ├── test-template.js      # Reusable test template
-    ├── playwright.config.ts  # Example config
+    ├── SKILL.md              # Main skill file (required)
+    ├── scripts/              # Executable code agents can run
+    │   └── test-template.js
+    ├── references/           # Additional documentation
+    │   └── REFERENCE.md
+    ├── assets/               # Static resources (templates, schemas)
+    │   └── config-template.json
     └── examples/
         ├── login-test.js
         └── api-test.js
 ```
 
+### Recommended directories (from agentskills.io spec)
+
+| Directory | Purpose |
+|-----------|---------|
+| `scripts/` | Executable code (Python, Bash, JavaScript). Should be self-contained with helpful error messages |
+| `references/` | Additional documentation loaded on demand (keep files focused for efficient context use) |
+| `assets/` | Static resources: templates, images, data files, schemas |
+
+### Progressive disclosure guidelines
+
+Skills load progressively to minimize context usage:
+
+1. **Metadata** (~100 tokens): `name` and `description` loaded at startup for all skills
+2. **Instructions** (< 5000 tokens recommended): Full SKILL.md body loaded when activated
+3. **Resources** (as needed): Files in `scripts/`, `references/`, `assets/` loaded only when required
+
+Keep your main SKILL.md under **500 lines**. Move detailed reference material to separate files.
+
 ## Reference documentation
 
 - [Agent Skills in VS Code](https://code.visualstudio.com/docs/copilot/customization/agent-skills)

.github/skills/copilot-skill-builder/skill-template.md

@@ -4,6 +4,12 @@ description: {{skill-description}}
 # argument-hint: {{optional hint text for slash command}}
 # user-invocable: true
 # disable-model-invocation: false
+# license: {{optional license name or file reference}}
+# compatibility: {{optional environment requirements}}
+# metadata:
+#   author: {{optional author}}
+#   version: "1.0"
+# allowed-tools: {{optional space-delimited list of pre-approved tools (experimental)}}
 ---
 
 # {{Skill Title}}