Skip to content

The Integrate Prompt

The integrate prompt is your zero-to-working starting point for any repository. Open your editor or the Copilot CLI and say:

integrate and audit BaseCoat from https://github.com/YOUR-ORG/YOUR-REPO

That single sentence triggers a guided two-phase workflow: integration then audit. No docs-hunting, no guessing.

What it does

Phase 1 — Integrate

The prompt inspects the target repository (tech stack, existing Copilot customizations, CI workflows, team size signals) and then:

  1. Recommends the right sync path for the context — direct sync, fork-first, or org-wide rollout
  2. Generates the exact sync commands for the detected OS and shell
  3. Drafts a tailored .basecoat.yml with only the agents, skills, and instructions relevant to the stack — no noise
  4. Validates the sync completed correctly and diagnoses failures if it didn't

Phase 2 — Audit

After integration, it runs a baseline audit across five dimensions:

Dimension What it checks
Instruction coverage Global file size (token cost), scoped instructions, duplicate rules
Agent relevance Stack match, required sections, allowed_skills wiring
Skill coverage Skills present, frontmatter valid
CI integration Drift detection workflow, undefined secrets, test validation
Naming and taxonomy File naming conventions, deprecated terminology

Each finding is classified by severity (🔴 Critical / 🟠 High / 🟡 Medium / ⚪ Low) and comes with a specific recommended fix.

How to invoke it

The prompt works in any Copilot-enabled surface:

GitHub Copilot CLI

gh copilot suggest "integrate and audit BaseCoat from https://github.com/YOUR-ORG/YOUR-REPO"

VS Code Copilot Chat

Open the chat panel and type:

@workspace /integrate integrate and audit BaseCoat from https://github.com/YOUR-ORG/YOUR-REPO

Or, if the prompt file is synced to your repo, select it from the prompt picker (/) and pass the URL.

In your repo (already inside the target)

If you are already working inside the target repository, you can omit the URL:

integrate and audit BaseCoat

The prompt will use the codebase tool to inspect the current repo.

What you get back

A structured report with three sections:

1. Repo profile — detected tech stack, sync method chosen, assets overlaid, BaseCoat version pinned.

2. Audit findings table — every finding grouped by severity with a one-line fix for each.

3. Top 3 recommended next steps — highest-impact actions in priority order, each with exact instructions.

Then the prompt asks: "Would you like me to implement any of these now?"

Example output (abbreviated)

Repo: https://github.com/acme/payments-api
Tech stack: Python 3.12, FastAPI, PostgreSQL, GitHub Actions
Sync method: Direct sync (solo team, public repo)
Assets overlaid: 12 agents, 4 skills, 6 instructions
BaseCoat version: v3.25.0

AUDIT FINDINGS

🔴 Critical (1)
  [INSTR-001] Global instruction file is 5.2KB — adds ~5,200 tokens to every session.
              Fix: Extract Python-specific rules to a scoped instruction file (applyTo: "**/*.py").

🟠 High (2)
  [AGENT-001] No agent matches the detected FastAPI/Python stack.
              Fix: Add agents/fastapi-review.agent.md or use the code-review agent with Python examples.
  [CI-001]    No version drift detection workflow found.
              Fix: Call .github/workflows/check-basecoat-version-callable.yml from your CI pipeline.

🟡 Medium (1)
  [NAME-001]  agents/Deploy_Prod.agent.md uses PascalCase.
              Fix: Rename to agents/deploy-prod.agent.md (kebab-case required).

TOP 3 NEXT STEPS
1. Split the global instruction file — saves ~5K tokens per session
2. Add drift detection to CI — prevents silent BaseCoat rot
3. Rename the miscased agent file

The prompt file

The prompt is located at prompts/integrate.prompt.md in the BaseCoat repository. After running sync, it lands in your repo at .github/prompts/integrate.prompt.md and is immediately available in VS Code's prompt picker.