Marketplace Large

A large marketplace fixture with 6+ plugins, deep CLAUDE.md content, multiple hook event types, multiple MCP servers, and extensive shared configuration. Intended to produce a strictly larger total-estimated-tokens count than the small or medium fixtures.

Plugins

plugin-alpha — linting and static analysis
plugin-beta — code formatting
plugin-gamma — test runner and coverage
plugin-delta — release automation and tagging
plugin-epsilon — configuration auditing
plugin-zeta — security scanning
plugin-eta — documentation generation
plugin-theta — dependency management

Commands

npm run build — Build all workspace packages
npm test — Run the entire test suite
npm run lint — Run all linters
npm run format — Auto-format all source files
npm run release — Cut a new release
npm run audit — Security audit of dependencies
npm run docs:build — Build documentation site
npm run docs:serve — Serve documentation locally
npm run typecheck — Type-check without emitting
npm run clean — Remove build artifacts

Architecture

The marketplace hosts multiple plugins, each self-contained. Inter-plugin communication happens via well-defined contracts; no plugin imports from another directly. Shared primitives live in a workspace package consumed by every plugin.

Conventions

Conventional Commits with plugin scope: feat(plugin-name): description
Semantic versioning per plugin, coordinated via a release-please-style flow
Tests live alongside implementation files as *.test.mjs
Zero runtime npm dependencies in hooks and scanners where possible

Tooling

Node.js ≥ 18 (ES Modules, node:test)
TypeScript strict mode
ESLint + Prettier
Shared git hooks via hooks/hooks.json

Release Process

Each plugin version-bumps independently. The marketplace root README and per-plugin README are updated in the same commit as the version bump. A tag of the form {plugin}-v{semver} is pushed to Forgejo alongside the main branch commit.

Shared Patterns

Shared conventions and helper patterns are defined in the imported files below. Each import is a logical slice: one for naming conventions, one for error-handling patterns, one for test patterns.

@.claude/shared-naming.md @.claude/shared-errors.md @.claude/shared-tests.md

2.3 KiB Raw Blame History