Generate .github/copilot-instructions.md for a brownfield project. Reads discovery outputs and context to produce stack-specific, domain-aware AI instructions. Replaces any generic copilot-instructions.md.
Generate .github/copilot-instructions.md tailored to this specific brownfield project.
.discovery/context.json — stack, tools, architectureproject.json — name, type, domaindocs/discovery/l1-capabilities.md — L1 capabilitiesdocs/discovery/domain-model.md — entities, ownership, relationshipsdocs/discovery/blueprint-comparison.md — industry alignment (if present)Produce a .github/copilot-instructions.md with these sections:
One paragraph: what this codebase is, its domain, its primary purpose. Derived from project.json → domain and L1 capabilities.
Concrete, factual list of the detected stack — no guesses, only what's in .discovery/context.json:
From context.json → arch: style (monolith/layered/microservices), monorepo yes/no, service count. Include the main entrypoints if detected.
Key entities from docs/discovery/domain-model.md and their owning capabilities. List the top 5–10 entities with a one-line description. This anchors naming conventions.
List of L1 capabilities from docs/discovery/l1-capabilities.md. One line each: name + one-sentence description. This tells the AI what the codebase does.
Derive from the detected stack:
Three to five hard constraints for this specific stack:
Write valid Markdown. Be specific — use the actual project name, actual entity names, actual tool names throughout. Do not write generic placeholder text.
Print:
✔ .github/copilot-instructions.md written ({N} sections)