
Pysmith
- 1 repo stars
- Updated March 26, 2026
- jugrajsingh/skillgarden
Scaffold Python projects, manage configuration, and audit best practices with uv-native tooling.
About
PySmith handles Python project scaffolding, configuration management, and best-practices auditing using uv-native tooling. It bootstraps well-structured Python projects quickly. Best for starting or standardizing Python codebases.
- Python project scaffolding
- Config management
- uv-native tooling
Pysmith by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add jugrajsingh/skillgarden/plugin install pysmith@skillgardenAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| repo stars | ★ 1 |
|---|---|
| Last updated | March 26, 2026 |
| Repository | jugrajsingh/skillgarden ↗ |
What it does
Scaffold Python projects, manage configuration, and audit best practices with uv-native tooling.
README.md
SkillGarden
A plugin marketplace for Claude Code that brings developer best practices as composable plugins. Install what you need, they work together seamlessly.
Available Plugins
gitmastery
Git workflow validation with enforcement. Explicit file staging, conventional commits, no AI footers.
What it enforces:
git add .→ Blocked. List files explicitly.git commit -m "fixed stuff"→ Blocked. Usetype(scope): subject- AI footers → Blocked automatically
Commands:
| Command | Description |
|---|---|
/gitmastery:commit |
Atomic commits with conventional format |
/gitmastery:feature |
Start a feature branch |
/gitmastery:finish |
Complete current branch |
/gitmastery:release |
Create versioned release |
/gitmastery:changelog |
Generate Keep a Changelog format |
/gitmastery:init |
Initialize git-flow |
claudemd
CLAUDE.md lifecycle management. Generate, audit, sync, and optimize hierarchical context files throughout a repository.
Commands:
| Command | Description |
|---|---|
/claudemd:init |
Generate root + module-level CLAUDE.md files |
/claudemd:audit |
Health check all context files for anti-patterns and staleness |
/claudemd:sync |
Detect and fix drift between docs and codebase |
/claudemd:optimize |
Reduce context cost while preserving signal |
unstuck
Structured error escalation protocol. 3-strike workflow when stuck on a problem: diagnose, pivot, escalate.
Commands:
| Command | Description |
|---|---|
/unstuck |
Start the escalation protocol |
planner
Design brainstorming, implementation planning with 3-file persistence, session recovery, and worktree isolation. Drives the front of the development workflow.
Commands:
| Command | Description |
|---|---|
/planner:brainstorm |
Explore design ideas through guided dialogue |
/planner:plan |
Create implementation plan with task decomposition and batching |
/planner:resume |
Recover session state via 5-Question Reboot Test |
/planner:handoff |
Generate session transfer document |
/planner:worktree |
Set up git worktree isolation for a branch |
Persistence: Creates docs/plans/{slug}-task_plan.md, findings.md, and progress.md for cross-session project memory.
researcher
Parallel sub-agent research with persistent reports. Dispatches locator, analyzer, and pattern-finder agents for evidence-backed codebase analysis.
Commands:
| Command | Description |
|---|---|
/researcher:research |
Research a codebase question with parallel sub-agents |
/researcher:dispatch |
Dispatch parallel agents for independent problems |
Agents: locator (haiku, fast file finding), analyzer (sonnet, data flow tracing), pattern-finder (sonnet, existing pattern discovery).
tidyup
Codebase cleanup and hygiene. Assess dead code, duplication, and staleness, then execute cleanup with safety gates.
Commands:
| Command | Description |
|---|---|
/tidyup:assess |
Analyze codebase for cleanup candidates (read-only) |
/tidyup:cleanup |
Execute cleanup with user approval and test verification |
Safety: Never deletes files -- archives to .archive/ instead. Always runs tests after changes.
shipit
Batch execution with subagent pipeline, TDD enforcement, code review, verification, and branch completion. The execution engine for plans created by planner.
Commands:
| Command | Description |
|---|---|
/shipit:execute |
Execute plan with implement-review-verify pipeline |
/shipit:tdd |
Run RED-GREEN-REFACTOR cycle for a single task |
/shipit:review |
Dispatch comprehensive code review |
/shipit:verify |
Evidence-before-claims verification gate |
/shipit:ship |
Complete branch (merge, PR, keep, or discard) |
/shipit:describe-pr |
Generate PR description from changes and plan |
Pipeline: Each task goes through implementer → spec-reviewer → quality-reviewer agents. Uses Claude Code native Tasks for real-time coordination.
pysmith
Python project scaffolding and best practices. Generate pyproject.toml, Pydantic Settings, pre-commit hooks, and audit existing projects.
Commands:
| Command | Description |
|---|---|
/pysmith:setup |
Full Python dev environment (pyproject + settings + pre-commit + makefile) |
/pysmith:pyproject |
Generate pyproject.toml with uv-native config |
/pysmith:settings |
Generate Pydantic Settings + YAML configuration |
/pysmith:precommit |
Generate .pre-commit-config.yaml with security hooks |
/pysmith:patterns |
Browse 9 copy-paste Python code patterns |
/pysmith:audit |
Audit project against Python best practices |
makesmith
Makefile generation for development and deployment. Supports role-separated Makefiles with self-documenting targets.
Commands:
| Command | Description |
|---|---|
/makesmith:local |
Generate Makefile.local with dev targets (venv, test, lint) |
/makesmith:deploy |
Generate Makefile.deploy with DevOps targets (build, push, deploy) |
/makesmith:makefile |
Generate root Makefile that delegates to local/deploy |
/makesmith:audit |
Audit Makefiles against conventions |
/makesmith:precommit |
Generate mbake pre-commit hook |
dockercraft
Docker best practices with language-aware generation. Supports Python, Node.js, Go, Rust, Java with multi-stage builds, compose orchestration, and security auditing.
Commands:
| Command | Description |
|---|---|
/dockercraft:setup |
Full Docker environment (Dockerfile + compose + services) |
/dockercraft:dockerfile |
Generate optimized Dockerfile with language detection |
/dockercraft:compose |
Generate docker-compose.yml with service detection |
/dockercraft:audit |
Audit Dockerfile and compose against security checklist |
/dockercraft:optimize |
Analyze and optimize Docker image size |
/dockercraft:precommit |
Generate hadolint pre-commit hook |
Agent: dockerfile-reviewer (sonnet, security + performance review with severity grading).
grepai
GrepAI semantic code search setup, initialization, and health monitoring. Guided orchestrator for Docker infrastructure, embedding provider/model selection, storage backend, and MCP registration.
Commands:
| Command | Description |
|---|---|
/grepai:setup |
Full guided setup (Docker, embeddings, storage, MCP, indexing) |
/grepai:init |
Initialize grepai in current project with provider/model/storage config |
/grepai:status |
Health check all components (Docker, Ollama, PostgreSQL, index, MCP) |
Infrastructure: Includes docker-compose.yml template with Ollama + PostgreSQL/pgvector.
branch-guardian
Git-flow branch protection. Blocks direct commits to main/develop.
Cross-Plugin Workflows
Planning and Execution
Idea -> /planner:brainstorm -> design doc
-> /planner:plan -> task plan + persistence files
-> /researcher:research -> evidence-backed analysis (optional)
-> /shipit:execute -> batch execution with review pipeline
-> /shipit:verify -> evidence gate
-> /shipit:review -> comprehensive code review
-> /shipit:ship -> merge, PR, keep, or discard
Python Project Setup
/pysmith:setup
-> generates pyproject.toml, config/settings.py, .pre-commit-config.yaml
-> delegates to /makesmith:local for Makefile.local
-> runs make -f Makefile.local setup-local
/dockercraft:setup
-> generates Dockerfile, .dockerignore
-> generates docker-compose.yml
-> optionally starts services
/makesmith:deploy
-> generates Makefile.deploy with Docker build/push/deploy
-> references Dockerfile built by dockercraft
Session recovery: /planner:resume reads persistence files and rebuilds context.
Session transfer: /planner:handoff generates a document for the next session.
Maintenance: /tidyup:cleanup assesses and removes dead code with safety gates.
Installation
Add the marketplace:
/plugin marketplace add jugrajsingh/skillgarden
Install plugins you need:
/plugin install gitmastery@skillgarden
/plugin install claudemd@skillgarden
/plugin install unstuck@skillgarden
/plugin install planner@skillgarden
/plugin install researcher@skillgarden
/plugin install tidyup@skillgarden
/plugin install shipit@skillgarden
/plugin install pysmith@skillgarden
/plugin install makesmith@skillgarden
/plugin install dockercraft@skillgarden
/plugin install grepai@skillgarden
Verify installation:
/help
# Should see plugin commands listed
Quick Start
After installing gitmastery, just work normally:
# This gets blocked:
git add .
# → "List files explicitly."
# This works:
git add src/auth.py src/utils.py
# Or use the guided workflow:
/gitmastery:commit
# → Analyzes changes, groups logically, presents each commit for approval
Philosophy
- Enforce, don't suggest - Block bad patterns before they happen
- Guide, don't lecture - Clear error messages with the fix
- Compose, don't monolith - Small plugins that do one thing well
- Persist, don't forget - 3-file persistence for cross-session memory
- Evidence, don't claim - file:line references and test proof required
Semantic Code Search with grepai
SkillGarden configures Claude Code to use grepai as its primary code exploration tool. grepai is a privacy-first semantic search engine that indexes the meaning of your code using vector embeddings, enabling natural language queries instead of exact text matching.
Why grepai?
| Standard Grep | grepai |
|---|---|
| Exact text matching | Understands intent |
| Need to know variable names | Search by what code does |
| No call graph awareness | Trace callers/callees |
grep "authenticate" |
grepai search "user login flow" |
Setup
Use the guided setup command:
/grepai:setup
This walks through Docker infrastructure, embedding provider/model selection, storage backend, MCP registration, and indexing — all interactively.
Or set up manually:
# Start Ollama (embeddings)
docker compose up -d
# Pull the embedding model
docker exec ollama ollama pull nomic-embed-text
# Initialize grepai in your project
grepai init
The default config uses GOB (local file) storage and nomic-embed-text embeddings — no PostgreSQL needed:
embedder:
provider: ollama
model: nomic-embed-text
endpoint: http://localhost:11434
store:
backend: gob
Start the file watcher to build and maintain the index:
grepai watch
Check health of all components:
/grepai:status
How Claude Code Uses It
CLAUDE.md configures Claude to prefer grepai over standard Grep/Glob for exploration:
- Semantic search —
grepai search "plugin validation logic"finds relevant code even without exact keyword matches - Call graph tracing —
grepai trace callers "HandleRequest"shows all callers before modifying a function - Fallback — Standard Grep/Glob for exact text matching (variable names, imports, specific strings)
MCP Integration
grepai exposes an MCP server that Claude Code connects to automatically, providing grepai_search, grepai_trace_callers, grepai_trace_callees, and grepai_trace_graph as native tools.
Contributing
Want to add a plugin? See CONTRIBUTING.md for development setup and guidelines.
Updating
/plugin update gitmastery@skillgarden
Or update all:
/plugin marketplace update skillgarden
Support
- Issues: https://github.com/jugrajsingh/skillgarden/issues
- Author: Jugraj Singh (jugrajskhalsa@gmail.com)