Claude Code Kit
A curated collection of agents, skills, and settings that supercharge Claude Code. The goal of this kit is to gather battle-tested subagents, reusable skills, and sensible default settings in one place so practitioners can drop them into their local Claude Code environment and immediately boost productivity.
🚀 What's Inside
| Component |
Count |
Location |
Installs to |
| Agents |
133 specialized subagents |
agents/ |
~/.claude/agents/ |
| Skills |
197 reusable skills |
skills/ |
~/.claude/skills/ |
| Settings |
Global config + permissions |
settings/ |
~/.claude/ |
| Hooks |
4 automation hooks |
hooks/ |
~/.claude/hooks/ |
- Agents are role-focused subagents (language experts, reviewers, architects, etc.) that Claude Code can delegate work to.
- Skills are self-contained capability packages (each with a
SKILL.md) that teach Claude how to use a specific tool, library, or workflow.
- Settings provide a ready-to-use
settings.json with environment tuning, permission guardrails, and a status line.
🔧 Installation (user-level / global)
These steps install everything at the user level so the agents, skills, and settings are available across all your projects (Claude Code reads from ~/.claude/ on Linux/macOS and %USERPROFILE%\.claude\ on Windows).
Run the commands from the root of this repository.
Linux / macOS
Dependencies: node ≥ 18 (for hooks) and jq (for the status line). Both are optional if you skip hooks or the status line.
# Debian/Ubuntu
sudo apt install nodejs jq
# macOS
brew install node jq
# 1. Create the target directories if they don't exist
mkdir -p ~/.claude/agents ~/.claude/skills ~/.claude/hooks/pre-tool-use ~/.claude/hooks/post-tool-use ~/.claude/hooks/notification
# 2. Copy all agents
cp -v agents/*.md ~/.claude/agents/
# 3. Copy all skills (each skill is its own directory)
cp -rv skills/* ~/.claude/skills/
# 4. Copy hooks
cp -v hooks/pre-tool-use/*.js ~/.claude/hooks/pre-tool-use/
cp -v hooks/post-tool-use/*.js ~/.claude/hooks/post-tool-use/
cp -v hooks/notification/*.js ~/.claude/hooks/notification/
# 5. Copy the status line script and make it executable
cp settings/statusline.sh ~/.claude/statusline.sh
chmod +x ~/.claude/statusline.sh
# 6. Copy the platform settings (already points to ~/.claude/statusline.sh)
cp -v settings/settings-bash.json ~/.claude/settings.json
Re-syncing / updates (recommended)
rsync is idempotent — re-run it anytime to pick up updates without clobbering unrelated files:
rsync -av agents/ ~/.claude/agents/
rsync -av skills/ ~/.claude/skills/
rsync -av hooks/ ~/.claude/hooks/
cp settings/statusline.sh ~/.claude/statusline.sh && chmod +x ~/.claude/statusline.sh
rsync -av settings/settings-bash.json ~/.claude/settings.json
Requires jq (brew install jq / apt install jq). git is optional — the script degrades gracefully when not in a repo.
Windows (PowerShell)
Dependencies: node ≥ 18 (for hooks). Install via nodejs.org or:
winget install OpenJS.NodeJS
# 1. Create the target directories if they don't exist
New-Item -ItemType Directory -Force `
"$env:USERPROFILE\.claude\agents", `
"$env:USERPROFILE\.claude\skills", `
"$env:USERPROFILE\.claude\hooks\pre-tool-use", `
"$env:USERPROFILE\.claude\hooks\post-tool-use", `
"$env:USERPROFILE\.claude\hooks\notification"
# 2. Copy all agents
Copy-Item agents\*.md "$env:USERPROFILE\.claude\agents\"
# 3. Copy all skills (each skill is its own directory)
Copy-Item skills\* "$env:USERPROFILE\.claude\skills\" -Recurse
# 4. Copy hooks
Copy-Item hooks\pre-tool-use\*.js "$env:USERPROFILE\.claude\hooks\pre-tool-use\"
Copy-Item hooks\post-tool-use\*.js "$env:USERPROFILE\.claude\hooks\post-tool-use\"
Copy-Item hooks\notification\*.js "$env:USERPROFILE\.claude\hooks\notification\"
# 5. Copy the status line script
Copy-Item settings\statusline.ps1 "$env:USERPROFILE\.claude\statusline.ps1"
# 6. Copy the platform settings and inject your username into path placeholders
$settings = Get-Content settings\settings-powershell.json -Raw
$settings = $settings -replace '<YourName>', $env:USERNAME
$settings | Set-Content "$env:USERPROFILE\.claude\settings.json" -Encoding utf8
Re-syncing / updates (recommended)
robocopy is the Windows equivalent of rsync — safe to re-run, only copies changed files:
robocopy agents "$env:USERPROFILE\.claude\agents" *.md /NFL /NDL
robocopy skills "$env:USERPROFILE\.claude\skills" /E /NFL /NDL
robocopy hooks "$env:USERPROFILE\.claude\hooks" /E /NFL /NDL
Copy-Item settings\statusline.ps1 "$env:USERPROFILE\.claude\statusline.ps1" -Force
$settings = Get-Content settings\settings-powershell.json -Raw
$settings = $settings -replace '<YourName>', $env:USERNAME
$settings | Set-Content "$env:USERPROFILE\.claude\settings.json" -Encoding utf8
If you have PowerShell 7 installed, open ~/.claude/settings.json after install and swap powershell for pwsh in the statusLine.command value. Windows Terminal is recommended for correct ANSI / UTF-8 rendering. No jq required — the script uses PowerShell's native JSON parser.
Heads up — settings. Copying settings.json overwrites your existing global Claude Code settings. If you already have a customized file, back it up first:
- Linux/macOS:
cp ~/.claude/settings.json ~/.claude/settings.json.bak
- Windows:
Copy-Item "$env:USERPROFILE\.claude\settings.json" "$env:USERPROFILE\.claude\settings.json.bak"
Then merge by hand instead of overwriting.
Note. The agents/ folder also contains a CLAUDE.md and an AGENTS-REFERENCE.md. These are reference/instruction docs, not agents — you can skip copying them, or copy them deliberately if you want their guidance.
After installing, restart Claude Code (or start a new session). The agents become available for delegation, and skills trigger automatically based on their descriptions.
🤖 Agents
133 subagents grouped by domain. Each links to its definition file.
Meta / Orchestration
| Agent |
Description |
| grand-architect |
Meta-orchestrator that plans complex features, breaks down large tasks, and coordinates multiple agents. |
| tech-lead-orchestrator |
Senior tech lead that analyzes projects and returns structured task breakdowns for agent coordination. |
| project-analyst |
Analyzes unfamiliar codebases to detect frameworks, tech stacks, and architecture for routing. |
| code-archaeologist |
Explores and documents unfamiliar, legacy, or complex codebases with full risk/action reports. |
Languages & Runtimes
| Agent |
Description |
| python-pro |
Type-safe, production-ready Python with modern async patterns and extensive typing. |
| typescript-pro |
Advanced TypeScript type system, full-stack development, and build optimization. |
| javascript-pro |
Modern ES2023+ JavaScript for browser, Node.js, and full-stack apps. |
| golang-pro |
Idiomatic Go for concurrent, high-performance, cloud-native systems. |
| rust-engineer |
Rust systems programming with memory safety and zero-cost abstractions. |
| cpp-pro |
High-performance modern C++20/23 and template metaprogramming. |
| java-pro |
Modern Java 21+ with virtual threads, pattern matching, and Spring Boot 3.x. |
| java-architect |
Enterprise Java architecture across the Spring ecosystem and microservices. |
| csharp-developer |
ASP.NET Core APIs and cloud-native .NET with clean architecture. |
| dotnet-core-expert |
Cloud-native .NET Core microservices with minimal APIs. |
| dotnet-framework-4.8-expert |
Legacy .NET Framework 4.8 maintenance and modernization. |
| php-pro |
Modern PHP 8.3+ with strong typing and enterprise frameworks. |
| kotlin-specialist |
Kotlin coroutines, multiplatform, and Android development. |
| swift-expert |
Native iOS/macOS/server-side Swift with advanced concurrency. |
| elixir-expert |
Fault-tolerant concurrent systems with OTP and Phoenix. |
| sql-pro |
Complex SQL query optimization and schema design across major engines. |
Frameworks (Frontend / Backend / Mobile)
| Agent |
Description |
| react-specialist |
React 18+ patterns, performance optimization, and server components. |
| react-coder |
Simplicity-first React 19 components using internal UI packages. |
| vue-expert |
Vue 3 Composition API, reactivity, and Nuxt 3 development. |
| angular-architect |
Enterprise Angular 15+ with complex state and micro-frontends. |
| nextjs-developer |
Production Next.js 14+ with App Router and server components. |
| frontend-developer |
Robust, scalable React frontend components and UX. |
| fullstack-developer |
End-to-end feature delivery from database to UI. |
| backend-developer |
Scalable API development and microservices. |
| django-developer |
Django 4+ apps and REST APIs with async views. |
| rails-expert |
Rails apps with Hotwire reactivity and idiomatic patterns. |
| laravel-specialist |
Laravel 10+ with Eloquent, queues, and enterprise features. |
| laravel-backend-expert |
Laravel backend controllers, services, and Eloquent models. |
| laravel-eloquent-expert |
Laravel Eloquent schemas, relationships, and query tuning. |
| spring-boot-engineer |
Enterprise Spring Boot 3+ microservices and reactive patterns. |
| wordpress-master |
WordPress themes, plugins, headless APIs, and scaling. |
| mobile-developer |
Cross-platform mobile with React Native and Flutter. |
| mobile-app-developer |
Native and cross-platform iOS/Android development. |
| flutter-expert |
Cross-platform Flutter 3+ with custom UI and state management. |
| electron-pro |
Electron desktop apps with native OS integration and distribution. |
| cli-developer |
Command-line tools with intuitive design and cross-platform support. |
Architecture & API Design
Data & Databases
| Agent |
Description |
| database-administrator |
High-availability DBs, performance tuning, and disaster recovery. |
| database-optimizer |
Query optimization, indexing, caching, and partitioning. |
| postgres-pro |
PostgreSQL performance, replication, and advanced features. |
| data-engineer |
Scalable data pipelines, ETL/ELT, and data infrastructure. |
| data-analyst |
Business data insights, dashboards, and statistical analysis. |
| data-scientist |
Statistical analysis, ML, and data storytelling. |
| data-researcher |
Discover, collect, and validate data from multiple sources. |
AI / ML
| Agent |
Description |
| ai-engineer |
AI system design, model implementation, and production deployment. |
| ml-engineer |
Production ML training pipelines, serving, and retraining. |
| machine-learning-engineer |
Model deployment, serving infrastructure, and edge inference. |
| mlops-engineer |
ML infrastructure, CI/CD, versioning, and experiment tracking. |
| llm-architect |
Production LLM systems, fine-tuning, RAG, and inference serving. |
| nlp-engineer |
NLP/NLU/NLG with transformer models and production pipelines. |
| prompt-engineer |
Design, optimize, and evaluate prompts for production LLMs. |
| mcp-developer |
Model Context Protocol server and client development. |
Infrastructure / DevOps / Cloud
| Agent |
Description |
| cloud-architect |
Multi-cloud (AWS/Azure/GCP) IaC, FinOps, and architecture. |
| devops-engineer |
CI/CD, containerization, monitoring, and automation. |
| deployment-engineer |
CI/CD pipelines and zero-downtime deployment strategies. |
| platform-engineer |
Internal developer platforms, golden paths, and self-service. |
| kubernetes-architect |
Cloud-native K8s, GitOps, service mesh, and multi-tenancy. |
| kubernetes-specialist |
Production-grade K8s deployments and cluster management. |
| docker-expert |
Container image building, optimization, and orchestration. |
| terraform-engineer |
Multi-cloud Terraform IaC and state management. |
| terragrunt-expert |
Terragrunt orchestration and DRY multi-environment IaC. |
| azure-infra-engineer |
Azure infrastructure, Entra ID, Bicep, and PowerShell automation. |
| network-engineer |
Cloud/hybrid network design, security, and troubleshooting. |
| cost-optimizer |
Cloud cost analysis, resource optimization, and scaling plans. |
| windows-infra-admin |
Windows Server, Active Directory, DNS, DHCP, and Group Policy. |
| iot-engineer |
Connected device architectures, edge computing, and IoT platforms. |
PowerShell / Windows Automation
Reliability / Operations / Incident Response
Security
Quality / Review / Testing
Performance & Build
UI / UX / Accessibility / Design
Documentation
Specialized Engineering
| Agent |
Description |
| fintech-engineer |
Financial systems, payments, and regulatory compliance. |
| slack-expert |
Slack apps, API integrations, and bot security review. |
Product / Business / Research
| Agent |
Description |
| product-manager |
Product strategy, feature prioritization, and roadmaps. |
| project-manager |
Project plans, risk management, and stakeholder coordination. |
| scrum-master |
Agile facilitation, ceremonies, and velocity improvement. |
| business-analyst |
Business process analysis and requirements gathering. |
| customer-success-manager |
Customer health, retention, and lifetime value. |
| sales-engineer |
Technical pre-sales, solution architecture, and POCs. |
| content-marketer |
SEO content strategy and multi-channel campaigns. |
| legal-advisor |
Contracts, compliance, IP strategy, and legal risk. |
| market-researcher |
Market analysis, consumer behavior, and opportunity sizing. |
| competitive-analyst |
Competitor analysis and competitive positioning. |
| research-analyst |
Multi-source research synthesis and trend identification. |
| trend-analyst |
Emerging patterns, industry shifts, and future scenarios. |
| search-specialist |
Advanced search strategies and targeted information retrieval. |
| scientific-literature-researcher |
Evidence-grounded answers from full-text research papers. |
🧩 Skills
197 skills, each packaged as a directory containing a SKILL.md (and supporting scripts/templates). Skills trigger automatically based on their descriptions. Grouped below by domain.
Token Efficiency / Caveman
| Skill |
Description |
| caveman |
Ultra-compressed communication mode (~75% token reduction). Lite / full / ultra / wenyan variants. Trigger: /caveman. |
| caveman-commit |
Terse Conventional Commits messages. ≤50-char subject, body only when "why" isn't obvious. Trigger: /caveman-commit. |
| caveman-compress |
Compress .md memory files to caveman prose (~46% input-token savings). Backs up originals. Trigger: /caveman-compress <file>. |
| caveman-help |
Quick-reference card for all caveman modes, skills, and commands. One-shot display. Trigger: /caveman-help. |
| caveman-review |
Ultra-compressed PR review: one line per finding — location, problem, fix. Trigger: /caveman-review. |
| caveman-stats |
Show real token usage and estimated savings for the current session via hook. Trigger: /caveman-stats. |
| cavecrew |
Decision guide for delegating to caveman-style subagents (~60% smaller tool results vs vanilla agents). |
Claude Code Workflow & Engineering Practice
| Skill |
Description |
| brainstorming |
Explore intent, requirements, and design before any creative/build work. |
| writing-plans |
Turn a spec into a multi-step implementation plan before coding. |
| executing-plans |
Execute a written plan in a separate session with review checkpoints. |
| subagent-driven-development |
Execute plans with independent tasks in the current session. |
| dispatching-parallel-agents |
Handle 2+ independent tasks with no shared state in parallel. |
| test-driven-development |
TDD before writing implementation code for features/bugfixes. |
| systematic-debugging |
Structured debugging before proposing fixes. |
| verification-before-completion |
Require verification commands before claiming work is done. |
| requesting-code-review |
Verify work meets requirements before merging. |
| receiving-code-review |
Rigorously evaluate review feedback before implementing. |
| finishing-a-development-branch |
Decide how to integrate completed work (merge/PR/cleanup). |
| using-git-worktrees |
Create isolated workspaces via git worktrees. |
| using-superpowers |
Establish how to find and use skills at conversation start. |
| writing-skills |
Create, edit, and verify skills before deployment. |
| skill-creator |
Create, improve, eval, and benchmark skills. |
| autoskill |
Observe workflows via screenpipe and draft new skills. |
| pi-agent |
Build with and use the Pi minimal terminal coding harness. |
| mcp-builder |
Build high-quality MCP servers (Python FastMCP / TS SDK). |
| claude-api |
Reference for the Claude API / Anthropic SDK: model IDs, pricing, params, streaming, tool use. |
Expo / React Native Development
| Skill |
Description |
| building-native-ui |
Build apps with Expo Router: styling, navigation, animations. |
| native-data-fetching |
Network requests, React Query/SWR, caching, and offline support. |
| add-app-clip |
Add an iOS App Clip target to an Expo app. |
| expo-api-routes |
Create API routes in Expo Router with EAS Hosting. |
| expo-brownfield |
Integrate Expo/React Native into existing native apps. |
| expo-cicd-workflows |
Author EAS workflow YAML for build/deploy pipelines. |
| expo-deployment |
Deploy Expo apps to App Store, Play Store, and web. |
| expo-dev-client |
Build and distribute Expo development clients. |
| expo-module |
Build Expo native modules/views (Swift, Kotlin, TS). |
| expo-observe |
Add and query EAS Observe performance metrics. |
| expo-tailwind-setup |
Set up Tailwind v4 / NativeWind v5 in Expo. |
| expo-ui-jetpack-compose |
Use Jetpack Compose views via @expo/ui. |
| expo-ui-swift-ui |
Use SwiftUI views via @expo/ui. |
| eas-update-insights |
Check EAS Update health: crash rates, installs, payload size. |
| upgrading-expo |
Upgrade Expo SDK versions and fix dependency issues. |
| use-dom |
Run web code in a webview via Expo DOM components. |
Web / Frontend / Design
Documents, Office & Publishing
| Skill |
Description |
| docx |
Create, read, and edit Word (.docx) documents. |
| pptx |
Create, read, and edit PowerPoint (.pptx) presentations. |
| xlsx |
Open, read, edit, and create spreadsheets (.xlsx/.csv/.tsv). |
| pdf |
Read, merge, split, fill, OCR, and create PDF files. |
| markitdown |
Convert office/media files to Markdown. |
| liteparse |
Local PDF/DOC parsing with bounding boxes and OCR for RAG. |
| doc-coauthoring |
Structured workflow for co-authoring documentation. |
| markdown-mermaid-writing |
Markdown + Mermaid diagram writing standards. |
| slides-generator |
Build animation-rich HTML presentations (or convert PPT). |
| internal-comms |
Write internal communications (status reports, updates, FAQs). |
Research, Search & Reasoning
Scientific Writing & Publishing
Data Science, Stats & Visualization
| Skill |
Description |
| exploratory-data-analysis |
EDA across 200+ scientific file formats with reports. |
| statistical-analysis |
Guided test selection, assumptions, and APA reporting. |
| statsmodels |
Statistical models (OLS, GLM, mixed, ARIMA) with diagnostics. |
| scikit-learn |
Supervised/unsupervised ML, pipelines, and tuning. |
| shap |
Model interpretability/explainability with SHAP. |
| umap-learn |
Nonlinear dimensionality reduction and embeddings. |
| matplotlib |
Low-level fully customizable plotting. |
| seaborn |
Statistical visualization with pandas integration. |
| polars |
High-performance DataFrame ETL/analytics. |
| dask |
Distributed/larger-than-RAM pandas/NumPy workflows. |
| vaex |
Out-of-core DataFrames for billions of rows. |
| networkx |
Create, analyze, and visualize graphs/networks. |
| sympy |
Exact symbolic math (algebra, calculus, solving). |
| matlab |
MATLAB/Octave numerical computing and visualization. |
| pymc |
Bayesian modeling and MCMC/variational inference. |
| pymoo |
Multi-objective optimization (NSGA-II/III, MOEA/D). |
| simpy |
Process-based discrete-event simulation. |
| aeon |
Time series ML (classification, forecasting, clustering). |
| timesfm-forecasting |
Zero-shot time series forecasting with TimesFM. |
| scikit-survival |
Survival analysis and time-to-event modeling. |
| usfiscaldata |
Query the U.S. Treasury Fiscal Data REST API. |
| get-available-resources |
Detect CPU/GPU/memory/disk and recommend a compute strategy. |
Deep Learning, RL & GPU
| Skill |
Description |
| transformers |
Hugging Face Transformers inference and fine-tuning. |
| pytorch-lightning |
Organize PyTorch with Lightning for scalable training. |
| torch-geometric |
Graph neural networks with PyTorch Geometric. |
| stable-baselines3 |
Standard RL algorithms with a scikit-learn-like API. |
| pufferlib |
High-performance parallel/multi-agent RL. |
| optimize-for-gpu |
GPU-accelerate Python (CuPy, Numba, cuDF, cuML, etc.). |
| modal |
Serverless cloud GPU compute with the Modal SDK. |
| hugging-science |
Discover scientific datasets/models/Spaces on Hugging Face. |
Bioinformatics & Genomics
| Skill |
Description |
| biopython |
Molecular biology toolkit: sequences, parsing, NCBI access. |
| bioservices |
Unified interface to 40+ bioinformatics services. |
| gget |
Fast CLI/Python queries to 20+ bioinformatics databases. |
| pysam |
Read/write SAM/BAM/CRAM/VCF/FASTA for NGS pipelines. |
| scanpy |
Standard single-cell RNA-seq analysis pipeline. |
| anndata |
Annotated matrix data structure for single-cell. |
| scvi-tools |
Deep generative models for single-cell omics. |
| scvelo |
RNA velocity analysis for single-cell trajectories. |
| cellxgene-census |
Query CZ CELLxGENE Census for single-cell data. |
| bulk-rnaseq |
End-to-end bulk RNA-seq orchestration (FASTQ to DEG). |
| pydeseq2 |
Differential gene expression for bulk RNA-seq. |
| pathway-enrichment |
Pathway/gene-set enrichment analysis (ORA/GSEA). |
| arboreto |
Infer gene regulatory networks (GRNBoost2, GENIE3). |
| geniml |
ML on genomic interval data (region embeddings, scATAC). |
| gtars |
High-performance genomic interval analysis (Rust). |
| polars-bio |
Genomic interval ops and bioinformatics I/O on Polars. |
| deeptools |
NGS analysis: bigWig, QC, heatmaps for ChIP/RNA/ATAC-seq. |
| pyopenms |
Mass spectrometry / proteomics analysis platform. |
| matchms |
Spectral similarity and compound ID for metabolomics. |
| flowio |
Parse FCS flow cytometry files. |
| scikit-bio |
Microbiome analysis: alignments, diversity, ordination. |
| phylogenetics |
Build/analyze phylogenetic trees (MAFFT, IQ-TREE, FastTree). |
| etetoolkit |
Phylogenetic tree manipulation and visualization (ETE). |
| cobrapy |
Constraint-based metabolic modeling (FBA/FVA). |
| esm |
ESM3/ESMC protein models and ESMFold folding. |
| glycoengineering |
Analyze/engineer protein glycosylation. |
| tiledbvcf |
Scalable genomic variant storage/query with TileDB. |
| lamindb |
Lineage-native lakehouse for biological datasets/models. |
| depmap |
Cancer Dependency Map gene/drug dependency queries. |
| primekg |
Query the Precision Medicine Knowledge Graph. |
| nextflow |
Build/run Nextflow and nf-core bioinformatics pipelines. |
| pacsomatic |
Operate nf-core/pacsomatic tumor-normal workflows. |
Cheminformatics & Drug Discovery
| Skill |
Description |
| rdkit |
Cheminformatics toolkit for fine-grained molecular control. |
| datamol |
Pythonic RDKit wrapper for standard drug discovery. |
| deepchem |
Molecular ML with featurizers and MoleculeNet datasets. |
| torchdrug |
PyTorch GNNs for molecules and proteins. |
| molfeat |
Molecular featurization for ML (100+ featurizers). |
| medchem |
Medicinal chemistry filters and compound triage. |
| pytdc |
Therapeutics Data Commons AI-ready drug datasets. |
| diffdock |
DiffDock molecular docking and pose prediction. |
| molecular-dynamics |
MD simulations with OpenMM and MDAnalysis. |
| rowan |
Cloud-native molecular modeling workflow platform. |
Physics, Astronomy, Materials, Quantum & Geospatial
| Skill |
Description |
| astropy |
Astronomy/astrophysics workflows with Astropy. |
| pymatgen |
Materials science: structures, phase diagrams, band structure. |
| fluidsim |
Computational fluid dynamics simulations. |
| qiskit |
IBM quantum computing framework. |
| cirq |
Google quantum computing framework. |
| pennylane |
Hardware-agnostic quantum ML with autodiff. |
| qutip |
Open quantum systems simulation. |
| zarr-python |
Chunked N-D arrays for cloud storage. |
| geomaster |
Remote sensing, GIS, and spatial ML. |
| geopandas |
Geospatial vector data analysis. |
Clinical, Medical & Healthcare
Lab Automation & Platform Integrations
⚙️ Settings
The settings/ directory contains platform-specific Claude Code configuration files:
All three files share the same base configuration and include a hooks.PreToolUse section that wires the two pre-tool-use hooks (see Hooks below).
env — environment variables that tune the Claude Code harness:
| Variable |
Value |
What it does |
CLAUDE_AUTOCOMPACT_PCT_OVERRIDE |
70 |
Triggers automatic context compaction once the context window reaches 70% usage (instead of the higher default), keeping more room before the limit. |
BASH_MAX_OUTPUT_LENGTH |
20000 |
Maximum number of characters captured from a Bash command's output before it is truncated. Raised so long command outputs aren't cut short. |
MAX_MCP_OUTPUT_TOKENS |
8000 |
Caps the number of tokens returned by MCP (Model Context Protocol) tool calls, preventing a single MCP response from flooding the context. |
CLAUDE_CODE_EFFORT_LEVEL |
medium |
Sets the default reasoning/effort level Claude applies to tasks (low / medium / high), balancing speed against thoroughness. |
Other keys:
permissions.deny — read guardrails that block sensitive/noisy paths (./secrets/**, node_modules, build/dist/coverage/.next, logs, *.log).statusLine — runs the platform-specific status line script (model, dir, git branch, context bar, cost, rate limits).theme — dark.model — sonnet.
Adjust model, theme, and the permission lists to match your own environment before relying on them.
A settings.local.json (machine-local permission overrides) is intentionally not tracked in this repo — it is git-ignored. Create your own at ~/.claude/settings.local.json if you need per-machine permission allowlists.
🪝 Hooks
The hooks/ directory contains Node.js scripts that Claude Code executes automatically at defined lifecycle events. Only the pre-tool-use hooks are wired into settings.json — the others (post-tool-use, notification) must be added manually if you want them.
Requirement: Node.js ≥ 18 must be on your PATH. The hooks use only built-in Node.js modules (fs, path, child_process) — no npm install needed.
Pre-Tool-Use hooks (active by default)
These run before Claude executes a tool call and can block the action entirely.
| Hook |
Matcher |
What it blocks |
block-dangerous-commands.js |
Bash |
Catastrophic shell commands: rm -rf ~, dd to disk, fork bombs, force-push to main, git reset --hard, chmod 777, env dumps, and more. Configurable via SAFETY_LEVEL (critical / high / strict). |
protect-secrets.js |
Read|Edit|Write|Bash |
Access to sensitive files (.env, SSH keys, AWS/GCP/Azure credentials, keystores, .netrc, .npmrc, etc.) and bash commands that expose or exfiltrate secrets. Same SAFETY_LEVEL knob. |
Post-Tool-Use hooks (optional)
Run after a tool call succeeds. Not wired into settings.json by default.
| Hook |
Matcher |
What it does |
auto-stage.js |
Edit|Write |
Automatically runs git add <file> after every file edit so git status always reflects exactly what Claude changed. |
Notification hooks (optional)
Run when Claude Code fires a notification event (permission prompts, idle, MCP dialogs). Not wired into settings.json by default.
| Hook |
What it does |
notify-permission.js |
Sends a Slack message when Claude needs user input. Set CCH_SLA_WEBHOOK to your Slack Incoming Webhook URL. |
Hook logs
All hooks write structured JSONL entries to ~/.claude/hooks-logs/YYYY-MM-DD.jsonl for audit and debugging.
📁 Repository Layout
claude-code-kit/
├── agents/ # 133 subagent .md definitions (+ CLAUDE.md, AGENTS-REFERENCE.md)
├── skills/ # 197 skill directories, each with a SKILL.md
├── hooks/
│ ├── pre-tool-use/ # block-dangerous-commands.js, protect-secrets.js
│ ├── post-tool-use/ # auto-stage.js
│ └── notification/ # notify-permission.js
├── settings/ # settings.json (global config)
└── README.md