basecoat

name	basecoat
title	Base Coat Router & Agent Discovery
description	Single entry point for 28+ agents across 6 disciplines—discover or delegate across Development, Architecture, Quality, DevOps, Process, and Meta
compatibility	["agent:*"]
metadata	{"domain":"framework","maturity":"production","audience":["all"]}
allowed-tools	["bash","curl","git"]

Base Coat Router

The front door to the entire Base Coat framework. This skill routes requests to the right agent across 6 disciplines: Development, Architecture, Quality, DevOps, Process, and Meta.

Two modes of operation:

• Discovery — browse, search, and learn about available agents
• Delegation — route a prompt directly to the right agent

---

Usage Modes

Discovery Mode

Use discovery when you want to explore what's available or need help picking the right agent.

Command	What It Does
/basecoat	Full categorized agent catalog
/basecoat development	Show only Development agents
/basecoat quality	Show only Quality agents
/basecoat help [agent-name]	Detailed usage card for one agent
/basecoat find "[search term]"	Fuzzy search across agent keywords

When showing discovery results, format them by category:

🔨 Development

Agent	Description	Invoke With
@backend-dev	APIs, services, business logic	/basecoat backend [prompt]
@frontend-dev	UI, components, accessibility	/basecoat frontend [prompt]
@middleware-dev	API gateways, integration, events	/basecoat middleware [prompt]
@data-tier	Schema, migrations, queries	/basecoat data [prompt]

🏗️ Architecture

Agent	Description	Invoke With
@solution-architect	System design, trade-offs, ADRs	/basecoat architect [prompt]
@api-designer	OpenAPI specs, API contracts	/basecoat api-design [prompt]
@ux-designer	Wireframes, accessibility, UX flows	/basecoat ux [prompt]

🔍 Quality

Agent	Description	Invoke With
@code-review	Code review, pull-request feedback	/basecoat review [prompt]
@security-analyst	Vulnerability scanning, threat models	/basecoat security [prompt]
@performance-analyst	Profiling, bottleneck analysis	/basecoat perf [prompt]
@config-auditor	Config hygiene, secrets audit	/basecoat config [prompt]
@manual-test-strategy	Test plans, manual test cases	/basecoat manual-test [prompt]
@exploratory-charter	Exploratory testing charters	/basecoat exploratory [prompt]
@strategy-to-automation	Convert test strategy to automation	/basecoat automate-tests [prompt]

🚀 DevOps

Agent	Description	Invoke With
@devops-engineer	CI/CD pipelines, infrastructure	/basecoat devops [prompt]
@release-manager	Versioning, changelogs, releases	/basecoat release [prompt]
@rollout-basecoat	Enterprise onboarding, rollout plans	/basecoat rollout [prompt]

📋 Process

Agent	Description	Invoke With
@sprint-planner	Sprint planning, wave breakdown	/basecoat sprint [prompt]
@product-manager	Requirements, user stories, PRDs	/basecoat product [prompt]
@issue-triage	Issue classification, labeling	/basecoat triage [prompt]
@retro-facilitator	Retrospective facilitation	/basecoat retro [prompt]
@project-onboarding	Project setup, getting-started guides	/basecoat onboarding [prompt]

🧰 Meta

Agent	Description	Invoke With
@agent-designer	Design new agents	/basecoat agent [prompt]
@prompt-engineer	System prompts, prompt tuning	/basecoat prompt [prompt]
@mcp-developer	MCP tool servers	/basecoat mcp [prompt]
@tech-writer	Docs, runbooks, ADRs	/basecoat docs [prompt]
@new-customization	Create new skills/customizations	/basecoat customization [prompt]
@merge-coordinator	Parallel merge conflict resolution	/basecoat merge [prompt]

---

Delegation Mode

When a discipline keyword and prompt are provided, the router delegates directly:

/basecoat backend build a REST API for user management
→ Delegates to @backend-dev with prompt "build a REST API for user management"

/basecoat review check auth module for SQL injection
→ Delegates to @security-analyst

/basecoat sprint plan sprint 12 from open issues
→ Delegates to @sprint-planner

---

Keyword-to-Agent Routing Table

Use this table to match user input to the correct agent.

Keyword(s)	Agent	Category
backend, api, server	@backend-dev	🔨 Development
frontend, ui, web	@frontend-dev	🔨 Development
middleware, integration, gateway	@middleware-dev	🔨 Development
data, database, db	@data-tier	🔨 Development
architect, design, system	@solution-architect	🏗️ Architecture
api-design, openapi, swagger	@api-designer	🏗️ Architecture
ux, accessibility, wireframe	@ux-designer	🏗️ Architecture
review, cr, pull-request	@code-review	🔍 Quality
security, vulnerability, threat	@security-analyst	🔍 Quality
perf, performance, profiling	@performance-analyst	🔍 Quality
config, secrets, audit	@config-auditor	🔍 Quality
manual-test, test-strategy	@manual-test-strategy	🔍 Quality
exploratory, charter	@exploratory-charter	🔍 Quality
automate-tests, test-automation	@strategy-to-automation	🔍 Quality
devops, cicd, deploy, pipeline	@devops-engineer	🚀 DevOps
release, version, changelog	@release-manager	🚀 DevOps
rollout, onboard-enterprise	@rollout-basecoat	🚀 DevOps
sprint, plan, wave	@sprint-planner	📋 Process
product, requirements, stories	@product-manager	📋 Process
triage, classify, label	@issue-triage	📋 Process
retro, retrospective	@retro-facilitator	📋 Process
onboarding, setup, getting-started	@project-onboarding	📋 Process
agent, create-agent	@agent-designer	🧰 Meta
prompt, system-prompt	@prompt-engineer	🧰 Meta
mcp, tools, tool-server	@mcp-developer	🧰 Meta
docs, document, runbook	@tech-writer	🧰 Meta
customization, create-skill	@new-customization	🧰 Meta
merge, conflict, parallel-merge	@merge-coordinator	🧰 Meta

---

Delegation Instructions

When routing a /basecoat [keyword] [prompt] request:

1. Match the first token after /basecoat against the keyword column in the routing table above.
2. Ambiguous match? If multiple agents could fit, show the top 2–3 candidates with a one-line description and ask the user to pick.
3. Load the agent — open the matched agent's instruction file and its paired skill from the skills/ directory.
4. Pass the prompt — forward everything after the keyword as the user's prompt to the loaded agent.
5. No match? Fall back to the full discovery menu with a note:

   I couldn't match "xyz" to a specific agent. Here's the full catalog — pick the one that fits:

---

Metadata Reference

The file basecoat-metadata.json at the repository root contains the machine-readable registry of all 28 agents. It includes:

• Agent names, descriptions, and file paths
• Category groupings
• Keywords and aliases for routing
• Paired skill references

The router can reference this file for programmatic matching when keyword lookup alone is insufficient.

---

Examples

# ── Discovery ──────────────────────────────────────────────
/basecoat                        → Full agent catalog by category
/basecoat architecture           → Architecture agents only
/basecoat help code-review       → Detailed usage card for @code-review
/basecoat find "deploy"          → Finds @devops-engineer, @release-manager

# ── Delegation ─────────────────────────────────────────────
/basecoat backend build a REST API for orders
/basecoat security run threat model for auth service
/basecoat sprint plan sprint 5 from open issues
/basecoat docs write a runbook for deployment
/basecoat agent design a new linting agent
/basecoat perf profile the checkout flow for latency