What does this skill do, and when should you use it?
This workflow skill comes from the Visual Studio Code Code - OSS repository and focuses on Agent customization files. It covers creating, updating, reviewing, fixing, and debugging .instructions.md, .prompt.md, .agent.md, SKILL.md, copilot-instructions.md, and AGENTS.md files. It also helps investigate why customizations are not loaded or invoked, including applyTo patterns, tool restrictions, custom agent modes, and YAML frontmatter. It is not intended for general coding questions, runtime debugging, MCP server configuration, or VS Code extension development.
Reads and writes Agent customization files; selects among agent instructions, File Instructions, Prompts, Hooks, Custom Agents, and Skills based on the request; asks questions to clarify requirements; uses read-only subagents for codebase exploration when needed; checks file locations, YAML frontmatter syntax, and the presence of a meaningful description; and explains the distinctions among applyTo patterns, tool restrictions, custom agents, prompts, and skills. For a single file created from a known pattern or a quick YAML fix, it recommends editing directly instead of invoking the skill.
- A developer wants project-wide coding preferences that apply consistently and needs to create or update copilot-instructions.md or AGENTS.md.
- A team member finds that an .instructions.md file is being ignored and needs to inspect its location, description, or applyTo pattern.
- A user wants a custom agent with tool restrictions or context isolation and needs to choose between a Custom Agent, Skill, and Prompt.
- A team wants to package domain knowledge as an on-demand workflow and needs guidance on organizing SKILL.md with references, scripts, or templates.
- A user encounters silent YAML frontmatter failures and needs to correct fields, indentation, quoting, or a name that does not match the folder.
How do you install this skill?
- The workflow involves writing customization files and potentially defining executable hooks; manually confirm target paths, tool permissions, command contents, and backup or rollback plans before use.
- Reference material is supplemented by external official links, so full validation may be unavailable offline or on restricted networks.
- Skill discovery and triggering depend on host-agent semantic recognition and may miss relevant requests or trigger on borderline cases.
- Local filesystem
The supplied source does not provide an installation command for the skill collection. The repository path for this skill is extensions/copilot/assets/prompts/skills/agent-customization/SKILL.md. The SKILL.md lists .github/skills/<name>/, .agents/skills/<name>/, and .claude/skills/<name>/ as skill locations; when using it elsewhere, follow the target Agent Skills client's installation procedure.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/microsoft/vscode.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/extensions/copilot/assets/prompts/skills/agent-customization" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Check why my .instructions.md is not being applied and fix its applyTo pattern.
- Create a custom agent for this project with restricted tools and valid YAML frontmatter.
Ask a compatible Agent for a concrete customization task, such as: “Check why my .instructions.md is not being applied and fix its applyTo pattern.” Another example is: “Create a custom agent for this project with restricted tools and valid YAML frontmatter.” If the task is only to create one file from a known pattern or make a quick YAML frontmatter fix, edit the file directly because the skill explicitly says those single operations do not need it.
What are this skill's strengths and limitations?
- Covers several VS Code Agent customization file types rather than focusing on only one.
- Provides a complete flow from determining scope and choosing a primitive to creating and validating the file.
- Clearly distinguishes the intended roles of Instructions, Skills, Prompts, Hooks, and Custom Agents.
- Targets concrete failure points such as applyTo, description, YAML frontmatter, and file placement.
- The source provides no standalone test suite, success-rate data, or compatibility test results.
- The source does not document a specific installation command for the skill collection.
- It depends on runtime access to filesystem tools, an ask-questions tool, and subagents; clients without those capabilities require adaptation.
- It explicitly excludes general coding questions, runtime error diagnosis, MCP configuration, and VS Code extension development.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| VS Code Agent Customization Assistant this page ✓ Microsoft · Official | 55 · Use with care | ★ 194k | 11d ago | MIT |
| Reusable Prompt Creator ✓ Microsoft · Official | 45 · Not recommended | ★ 194k | 11d ago | MIT |
| Codebase Customization Initializer ✓ Microsoft · Official | 45 · Not recommended | ★ 194k | 11d ago | MIT |
| SkillCheck Free | 64 · Recommended | ★ 41 | 14d ago | MIT |
| Claude Code Agent Development Guide ✓ Anthropic · Official | 54 · Use with care | ★ 38k | 11d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill defines scope boundaries, recommends read-only subagent exploration, and asks whether the target is workspace or user-level; however, its creation workflow still permits direct file writes without explicit per-action confirmation, sensitive-data-flow guidance, rollback, or detailed external-tool permission controls, so points are deducted.
The decision table, creation steps, boundaries, and pitfalls are broadly consistent and include basic validation requirements; however, referenced documents and tools depend on the host environment, while abnormal-input handling and diagnosable failure feedback are thin. Static review also caps this score below 10.
The description provides concrete trigger scenarios and non-fit cases across several VS Code customization files; however, inputs, expected outputs, semantic trigger boundaries, Chinese-language support, and mainland-China network fit are not sufficiently specified, so points are deducted.
The documentation uses a layered main document plus reference files and includes templates, locations, naming guidance, progressive loading, license context, and official-source pointers; however, the skill lacks its own versioning, changelog, explicit maintenance owner, and update path, with dependencies on external documentation.
It can guide primitive selection, file creation, and basic validation, so it addresses the core task; however, no representative output or execution result is verified, and practical completion depends on the host agent and unavailable tools. Static calibration limits this score to at most 7.
The material contains auditable templates and rules, plus repository CI and test context, but those workflows and tests do not cover this skill's key paths. There is no third-party execution evidence or independent reproduction, so only limited credit is warranted.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →