Claude Code Plugins

Community-maintained marketplace

Feedback

cli-ninja-tools

@killerapp/mem8-plugin
1
0

CLI power tools for AI-assisted development. Use when (1) needing recommendations for CLI tools to install, (2) processing JSON/YAML data with jq/yq, (3) searching code with ripgrep or ast-grep, (4) documenting a CLI tool or multi-tool recipe you've discovered, (5) wanting to learn CLI patterns for data pipelines, or (6) setting up a new project and want CLI recommendations. Supports three modes - init (project scan), document (capture new recipes), and recommend (codebase analysis).

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name cli-ninja-tools
description CLI power tools for AI-assisted development. Use when (1) needing recommendations for CLI tools to install, (2) processing JSON/YAML data with jq/yq, (3) searching code with ripgrep or ast-grep, (4) documenting a CLI tool or multi-tool recipe you've discovered, (5) wanting to learn CLI patterns for data pipelines, or (6) setting up a new project and want CLI recommendations. Supports three modes - init (project scan), document (capture new recipes), and recommend (codebase analysis).

CLI Ninja Tools

Transform terminal workflows with modern CLI utilities optimized for AI-assisted development.

Content Organization

CLI References (clis/)

Per-tool documentation with progressive disclosure:

  • clis/_index.md - Tier 1/2/3 tool index
  • clis/{tool}/quick.md - Essential patterns (~50 lines, always safe to load)
  • clis/{tool}/reference.md - Full documentation (load on demand)

Recipes (recipes/)

Multi-CLI combinations (2+ tools) for specific use cases:

  • recipes/_index.md - Recipe category index
  • recipes/{category}/*.md - Individual recipes (30-80 lines each)

Categories: github, data-processing, code-analysis, devops, media

Three Operational Modes

Mode A: Init Integration

When: Setting up new project, after /init Purpose: Scan project, recommend relevant CLI patterns

  1. Run python scripts/scan.py to detect CLIs in project configs
  2. Run python scripts/discover.py --missing to find missing Tier 1 tools
  3. Recommend patterns from recipes/ matching detected workflows
  4. Optionally add "## CLI Tricks" section to CLAUDE.md

Mode B: Documentation Helper

When: User discovers useful CLI pattern or recipe Purpose: Help capture terse, high-quality documentation

For single CLI:

  1. Run {cli} --help, optionally WebSearch for docs
  2. Generate clis/{cli}/quick.md following template
  3. Offer to save to skill or project CLAUDE.md

For recipe (multi-CLI):

  1. User provides command pipeline
  2. Decompose each component, explain data flow
  3. Generate recipe following recipes/TEMPLATE.md
  4. Validate with python scripts/validate.py

Mode C: Codebase CLI Recommendations

When: User explicitly asks for CLI recommendations Purpose: Analyze codebase patterns, recommend tools

  1. Run python scripts/discover.py for current tool state
  2. Run python scripts/scan.py for project patterns
  3. Correlate with skill repertoire
  4. Generate prioritized recommendations with install commands

Quick Start

Check Available Tools

python scripts/discover.py              # Full report
python scripts/discover.py --tier 1     # Essentials only
python scripts/discover.py --missing    # What to install

Essential Tools (Tier 1)

Tool Check Purpose
jq jq --version JSON processing
ripgrep rg --version Fast code search
fd fd --version File finder
bat bat --version Cat with syntax
ast-grep sg --version AST refactoring
# Quick check all Tier 1
for cmd in jq rg fd bat sg; do command -v $cmd &>/dev/null && echo "OK $cmd" || echo "MISSING $cmd"; done

Quick Install

# macOS
brew install jq ripgrep fd bat ast-grep

# Ubuntu/Debian
sudo apt install jq ripgrep fd-find bat
cargo install ast-grep --locked

# Windows (Scoop)
scoop install jq ripgrep fd bat ast-grep

Most Common Patterns

jq (JSON)

jq '.data[].name' response.json                    # Extract field
jq '.items[] | select(.active)' data.json          # Filter
jq -r '.users[] | [.id, .name] | @csv' users.json  # To CSV
jq -s '.[0] * .[1]' base.json override.json        # Merge

ripgrep (Search)

rg 'TODO|FIXME' -A 2 --type py                     # With context
rg 'pattern' --json | jq 'select(.type=="match")' # Structured output
rg -l 'deprecated' | xargs sd -i 'old' 'new'       # Search & replace

gh + jq (GitHub)

gh issue view 123 --json body --jq '.body'         # Inline jq
gh pr view 123 --json files --jq '.files[].path'   # PR files
gh issue list --json number,title --jq '.[].title' # List issues

Loading Strategy

When user asks about CLI tools:

  1. Start with this SKILL.md (overview)
  2. Load clis/{tool}/quick.md for specific tool questions
  3. Load clis/{tool}/reference.md only for advanced usage
  4. Load recipes/{category}/*.md for multi-tool workflows

Extending This Skill

Add Personal Recipes

Create ~/.claude/skills/cli-ninja-local/:

cli-ninja-local/
├── SKILL.md        # References this skill
├── recipes/        # Personal recipes
└── clis/           # Personal CLI docs

Contributing Back

  1. Create recipe following recipes/TEMPLATE.md
  2. Validate: python scripts/validate.py your-recipe.md
  3. PR to mem8-plugin

See recipes/CONTRIBUTING.md for details.

Reference Documentation