| name | chipyard |
| description | Unified Chipyard flow: automatically routes to local or Docker workflow based on backend_mode setting. |
| allowed-tools | ["Bash","Read","AskUserQuestion"] |
chip-agent:chipyard
Unified entry point. Routes to the appropriate workflow based on .chipagent/env.json configuration.
Mode Detection
PROJECT_ROOT="<project-root>"
BACKEND_MODE=$(grep -o '"backend_mode": *"[^"]*"' "$PROJECT_ROOT/.chipagent/env.json" | cut -d'"' -f4)
If backend_mode is missing or empty:
-
Run /chip-agent:initialize to detect installed tools and determine best backend mode
- This skill scans for local installations (Chipyard, ORFS, Java, sbt)
- It also checks if Docker is available
- Based on findings, it writes the optimal
backend_mode to .chipagent/env.json
-
Re-read backend_mode after initialization completes:
BACKEND_MODE=$(grep -o '"backend_mode": *"[^"]*"' "$PROJECT_ROOT/.chipagent/env.json" | cut -d'"' -f4)
-
Continue with routing using the newly detected backend_mode
If still empty after initialization, default to local.
Routing
| backend_mode | Skill to read | Description |
|---|
local | skills/chipyard-local/SKILL.md | Local Chipyard installation via MCP tools |
docker | skills/chipyard-docker/SKILL.md | Docker containers (ucbbar/chipyard-image) |
Execution: Read the appropriate skill file and follow its pipeline.
Configuration
Use either local or Docker MCP servers:
- Local:
mcp__chipagent-backend__* tools → run Chipyard locally
- Docker:
mcp__chipagent-backend-docker__* tools → run Chipyard in Docker containers
Then invoke this skill again.
