From anvil
Interactive project setup — detect tech stack, configure components, write docs/anvil/config.yml
How this skill is triggered — by the user, by Claude, or both
Slash command
/anvil:initThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Set up anvil for this project through interactive conversation.
Set up anvil for this project through interactive conversation.
Read docs/anvil/config.yml. If it exists, tell the user:
"Anvil is already configured for this project. Would you like to update the existing config or start fresh?"
If updating, read the current config and present it as the starting point.
Scan the project root for build/config files to detect components:
| File | Language | Likely Test Runner |
|---|---|---|
pyproject.toml, setup.py, setup.cfg | Python | pytest, unittest |
package.json | JavaScript/TypeScript | jest, vitest, mocha |
Cargo.toml | Rust | cargo test |
go.mod | Go | go test |
pom.xml, build.gradle | Java/Kotlin | JUnit, Maven Surefire |
Gemfile, *.gemspec | Ruby | RSpec, Minitest |
mix.exs | Elixir | ExUnit |
composer.json | PHP | PHPUnit |
*.csproj, *.sln | C#/.NET | dotnet test |
For each detected file, explore further:
package.json for scripts.test to determine test commandpyproject.toml for [tool.pytest] or [tool.ruff] sectionsShow the user what you found:
"I detected a Python project with pytest (
src/→test/) and a TypeScript frontend with vitest (apps/web/src/→apps/web/tests/). Sound right?"
For each component (detected or user-specified), confirm or ask:
api, frontend, worker)test/test_{module}.py)Ask:
mainfeature/Create docs/anvil/ directory if needed. Write docs/anvil/config.yml.
Check if ROADMAP.md exists at the project root:
/anvil:roadmap to create the project roadmap/anvil:sprint <phase> to create a sprint from an existing phasenpx claudepluginhub olino3/anvil --plugin anvilInteractively configures projects for TDD-based autonomous development: auto-detects tech stack from package.json/composer.json/etc., gathers test/lint commands, generates CLAUDE.md and .claude/ files.
Scans codebase, infers project configuration, and interactively generates SDLC files with confidence-driven questions.
Bootstraps new projects or improves existing ones with best practices for structure, git, documentation, testing, code quality, dependencies, dev workflow, and CI/CD.