| name | workspace-bootstrap-architect |
| description | Design or refine the ai-team customization and bootstrap layer for Copilot. Use when users want to create or improve `AGENTS.md`, `copilot-instructions.md`, file-specific instructions, custom agents, prompts, or overall customization layout while keeping `.ai-team/` as the source of truth and `.github/` as a thin compatibility layer. |
Workspace Bootstrap Architect
Primary fit: Emily Davis. Other agents may use this skill when the job is to shape the repository's Copilot-facing customization layer instead of editing product code.
What This Skill Is For
Use this skill to create or refine the repository's customization architecture:
AGENTS.md
.github/copilot-instructions.md
.ai-team/instructions/**/*
.ai-team/agents/**/*
.ai-team/prompts/**/*
.ai-team/skills/**/*
- optional thin compatibility artifacts in
.github/**/* when explicitly needed
Read These Sources First
AGENTS.md
.github/copilot-instructions.md
.ai-team/ai-team-way.md
analysis/copilot/copilot-files.md
analysis/copilot/copilot-project-setup-guide.md
.vscode/settings.json
- Existing
.ai-team/agents/**/*, .ai-team/skills/**/*, .ai-team/prompts/**/*, and .ai-team/instructions/**/*
.ai-team/instructions/agents.instructions.md
.ai-team/instructions/prompts.instructions.md
.ai-team/instructions/skills.instructions.md
Core stance
Treat:
.ai-team/ as the source of truth
.github/ as an optional bootstrap and compatibility layer
Do not duplicate large bodies of knowledge across both unless there is a compelling compatibility reason.
Workflow
1. Classify the artifact correctly
Decide whether the need belongs in:
- instructions → always-on or file-targeted policy
- prompt → focused human-launched task starter
- skill → repeatable on-demand workflow
- agent → reusable specialist teammate with stable role boundaries
- bootstrap doc → top-level discovery and navigation
2. Keep the layer thin and intentional
Before writing anything, ask:
- does this belong in
.ai-team/ or .github/?
- does the file reduce confusion, or just add another place for drift?
- is the guidance specific enough to be worth loading?
3. Preserve the ai-team way
When shaping customization files:
- agents should feel like focused coworkers
- in
.ai-team/agents/, each agent is a single .agent.md file with YAML frontmatter for all metadata and a Markdown body for the portfolio
- agents should greet naturally on their first reply when the conversation did not already begin with a greeting
- prompts should be personal, communicative, and scoped
- skills should stay procedural and auditable
- non-CEO agents should have explicit
reportsTo
- agent personality should suit the role
- YAML/frontmatter should only use fields that are supported and materially useful
4. Design for discovery
Descriptions and top sections should help Copilot route work correctly.
Prefer:
- concrete trigger phrases
- clear ownership
- small files with strong headings
- practical language over generic best-practice filler
5. Validate the layout
Before finishing, confirm:
- the file lives in the right layer
- it does not duplicate stronger source-of-truth content elsewhere
- it fits the configured discovery locations in
.vscode/settings.json
- the resulting customization set is easier to navigate than before
Working Rules
- prefer the smallest structural change that improves clarity
- avoid giant bootstrap docs that try to solve every future problem
- keep instructions artifact-specific when possible
- let Emily own role and customization shape; let John scout and shape skill capability inputs
- if GitHub compatibility is needed, mirror only the minimum necessary content and do not treat
.github/ as the default home for agents, prompts, or skills
Successful Outcome
- the repository's customization layer is easy to understand
.ai-team/ remains the true home for durable knowledge
.github/ stays thin and intentional
- agents, prompts, skills, and instructions each have a clean job