// Configure which review agents run for your project. Auto-detects Python stack and writes compound-engineering.local.md.
This skill should be used to refine brainstorm or plan documents before proceeding to the next workflow step. It applies when a brainstorm or plan document exists and the user wants to improve it.
Use when a workflow command needs to detect multi-service monorepos, prompt for service scope, and propagate `<focus_context>` to sub-agents. Triggers when `--services` is passed, when focus context already exists in the conversation, or when multiple `pyproject.toml` files exist at the repo root.
Browser automation using Vercel's agent-browser CLI. Use when you need to interact with web pages, fill forms, take screenshots, or scrape data. Alternative to Playwright MCP - uses Bash commands with ref-based element selection. Triggers on "browse website", "fill form", "click button", "take screenshot", "scrape page", "web automation".
Build applications where agents are first-class citizens. Use this skill when designing autonomous agents, creating MCP tools, implementing self-modifying systems, or building apps where features are outcomes achieved by agents operating in a loop.
Analyze changes, group into logical units, and create conventional commits with gitmoji. Use when committing code changes.
Implement async task queues with Celery and real-time WebSocket support with Django Channels. Use when a Django project includes celery or channels in its dependencies, or when implementing background processing, periodic tasks, WebSocket connections, or real-time features.
| name | setup |
| description | Configure which review agents run for your project. Auto-detects Python stack and writes compound-engineering.local.md. |
| model | sonnet |
| disable-model-invocation | true |
Interactive setup for compound-engineering.local.md — configures which agents run during /workflows:review and /workflows:work.
Detection rules, agent computation, workflow skill mapping, and output template: references/stack-detection.md.
Read compound-engineering.local.md in project root. If it exists, show current settings and AskUserQuestion:
question: "Settings file already exists. What would you like to do?"
header: "Config"
options:
- Reconfigure — run interactive setup again
- View current — show file contents, then stop
- Cancel — keep current settings
View current → display file, stop. Cancel → stop.
Scan dependency files (see references for full rules). Then AskUserQuestion:
question: "Detected Python project with: {detected_frameworks}. How would you like to configure?"
header: "Setup"
options:
- Auto-configure (Recommended) — smart defaults for your stack
- Customize — choose frameworks, focus areas, and depth
Auto-configure → skip to Step 4 using defaults (see references). Customize → Step 3.
Django / DRF / Celery / FastAPI — override detected set.
PostgreSQL / PostGIS / SQLAlchemy.
Security / Performance / Architecture / Code simplicity — maps to focus-area agents.
- Thorough (Recommended) — stack reviewers + all selected focus agents
- Fast — stack reviewers + code simplicity only
- Comprehensive — thorough + git history, data integrity, agent-native
- Auto-configure (Recommended) — load detected stack skills into all workflows
- Customize per workflow — multiSelect for each of 4 workflows
- Skip — omit workflow_skills entirely
If Customize per workflow, for each of brainstorm, plan, work, review, present multiSelect of available skills (django-patterns, django-drf, django-async, fastapi-patterns, fastapi-sqlalchemy, pytest-patterns, postgresql-patterns, postgis-patterns) plus any local .claude/skills/.
Compute review_agents and plan_review_agents from selections (algorithm in references/stack-detection.md).
Write compound-engineering.local.md using the template in references. Omit workflow_skills block if user chose Skip in 3e.
Saved to compound-engineering.local.md
Stack: Python
Frameworks: {framework list}
Database: {database list}
Review depth: {depth}
Agents: {count} configured
{agent list, one per line}
Workflow skills: {count} skills across {count} workflows
brainstorm: {skill list or "none"}
plan: {skill list or "none"}
work: {skill list or "none"}
review: {skill list or "none"}
Tip: Edit the "Review Context" section for project-specific instructions.
Edit workflow_skills in the YAML frontmatter to customize per-workflow.
Re-run this setup anytime.