Claude Code Tresor Toolkit
Add code review, test generation, and development workflow commands to Claude Code.
- Source repo
- alirezarezvani/claude-code-tresor
- Stars
- ★ 788
- Last updated
- 8mo ago
- License
- MIT
- Primary language
- Shell
- FA score
- 39/100 · Major gaps
At a glance
- How it runs
- Works with
- Platform-specificClaude Code
- Cost
- Free, no paid service needed
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A solo developer wants a React component scaffold in Claude Code and runs
/scaffold. - Not a fit if
- Teams needing native support outside Claude Code
- Website
- alirezarezvani.com ↗ set on GitHub
- Maintenance
- No commits for 6+ monthslast commit 2026-02-25latest release v2.6.0 (2025-11-16)
- Source review
- 39/100 · Major gaps
What does this agent do, and when should you use it?
Claude Code Tresor is a development toolkit for Claude Code users, combining automatically triggered skills, manually invoked agents, slash commands, prompt templates, development standards, and workflow examples. Its installer copies skills, agents, and commands into Claude Code’s configuration directories; the README also documents manual and component-selective installation. During use, skills run in the background, users invoke specialist agents with @agent, and start scaffolding, review, testing, documentation, or orchestration flows with slash commands. The repository lists 8 skills, 8 core agents, 133 extended subagents, and 19 commands. It suits individuals and teams looking to reuse development workflow utilities inside Claude Code; the README does not document direct use on other platforms.
The scripts/install.sh script installs skills, agents, and commands, with --skills-only, --commands-only, --agents-only, and --resources-only options for selective installation. Skills include code-reviewer, test-generator, security-auditor, secret-scanner, and dependency-auditor, with README-described triggers such as file edits, code types, or dependency changes. Users can request deeper analysis from @config-safety-reviewer, @test-engineer, or @security-auditor, and launch workflows with /scaffold, /review, /test-gen, /docs-gen, and 10 orchestration commands. The repository also provides 20+ prompt templates, development standards, and examples; the README does not describe a hosted service, API, or standalone server execution path.
- A solo developer wants a React component scaffold in Claude Code and runs
/scaffold. - A team wants to review staged or PR-scoped code and uses
/reviewwith security, performance, or configuration checks. - A test engineer needs tests for an existing file and uses
/test-genor invokes@test-engineer. - A security lead wants to check dependency vulnerabilities or code security and uses
/vulnerability-scanor@security-auditor. - A team wants consistent style, Git, and review practices and adopts the repository’s standards, templates, and examples.
How do you install or deploy this agent?
The README documents installation through a shell script. It installs skills, agents, and commands into Claude Code; the manual method copies them into the corresponding ~/.claude/ directories. The README does not list additional runtime or API-key requirements.
git clone https://github.com/alirezarezvani/claude-code-tresor.git
cd claude-code-tresor
chmod +x scripts/install.sh
./scripts/install.shTo install selected components, use the documented options:
./scripts/install.sh --skills-only
./scripts/install.sh --commands-only
./scripts/install.sh --agents-only
./scripts/install.sh --resources-onlyHow do you use this agent?
After installation, invoke slash commands or request an agent by its @ name in Claude Code. These examples are from the README:
/scaffold react-component UserProfile --hooks --tests
/review --scope staged --checks security,performance
/test-gen --file utils.js --coverage 90
/docs-gen api --format openapiYou can also request specialist analysis:
@config-safety-reviewer Review database connection pool settings for pool size limits and timeout configurations
@security-auditor Please review this component for OWASP Top 10 vulnerabilities and input validationWhat are this agent's strengths and limitations?
- Combines automatically triggered skills, manually invoked agents, and slash commands for distinct interaction patterns.
- The README lists 19 commands spanning scaffolding, review, tests, documentation, and security, performance, and operations orchestration.
- The installer supports selecting skills, agents, commands, or resources separately.
- The README positions the toolkit for Claude Code and does not document native installation or adapters for other platforms.
- Installation uses a shell script and local file copying; no graphical installer is documented.
- The README gives different counts and structures across version sections, so verify the published version’s actual component inventory before adopting it.
How does this agent compare with similar options?
The README presents Claude Code Skill Factory, for building custom skills and agents, and Claude Skills Library, for ready-made professional-domain skill packages, as related projects. Tresor is positioned as a collection of ready-to-use development workflow utilities.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Claude Code Tresor Toolkit This agent | 39 · Major gaps | Agent plugin / skillFree | ★ 788 | 8mo ago | Shell | Claude Code |
| Itqan Engineering | 70 · Some gaps | Agent plugin / skillFree + model costs | ★ 12 | 1mo ago | Shell | Codex · Claude Code |
| Agent Skills — Production-Grade Engineering Skills for AI Agents | 67 · Some gaps | CLIFree + model costs | ★ 119k | 2d ago | JavaScript | Codex · Claude Code · OpenAI API |
| Claude Forge | 55 · Major gaps | Agent plugin / skillFree + model costs | ★ 880 | 2mo ago | Shell | Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
The code-review skill declares only Read, Grep, and Glob and says it needs no network. The git-commit helper permits Bash, while test-generator and api-documenter permit writes; documentation also says skills have no Bash and broadly describes access as read-only. These conflicts support only partial least-privilege evidence. Examples mostly suggest or generate results, but do not establish a clear confirmation rule before writes, commits, or installation. The docs mention reading staged diffs and project files without systematically explaining data flow, external transmission, or retention. A secret-scanner is listed, but the supplied material does not show its implementation or credential-handling rules. CI pins some tools, but there is no evidence of a complete dependency inventory, lock strategy, or vulnerability response. Install/update scripts, Bash, and write-capable tools can cause external effects; their specific behavior, rollback, and recovery guarantees are not shown in the supplied excerpts. The README names an author and MIT license, but component-level provenance and attribution for third-party material are not established.
README, CLAUDE.md, installation guide, and FAQ conflict on component counts, names, and locations: 19 versus 4 commands, 133 versus 141 extended/total agents, and old and new names coexist. This materially weakens self-consistency. The CI excerpt pins some tooling and installation docs list Claude Code and Git prerequisites, but the supplied material lacks a complete dependency inventory or availability alternatives. Troubleshooting gives symptoms and steps for common problems, so failure guidance is partly addressed; several instructions depend on logs or configuration not shown here.
The material covers individual and team use across code review, testing, documentation, security, and operations, with examples in several languages. It explains the conceptual boundaries between skills, agents, and commands, but claims of automatic monitoring do not align cleanly with the supplied skill descriptions, triggers, permissions, and write capabilities. Trigger descriptions often use broad terms such as code, save, and commit; documented keyword and file-pattern configuration also differs from the supplied frontmatter, limiting evidence of trigger precision. Installation is centered on Claude Code CLI and ~/.claude paths, with little evidence of environment adaptation.
The repository structure, component categories, and cross-links are substantial, but README, CLAUDE.md, and guides conflict on counts, names, paths, and versions. Installation methods, prerequisites, selective install, troubleshooting, and uninstall steps are documented; however, component counts differ across guides, and the supplied packet does not include the installer. Naming stability is weak: the docs describe breaking renames while old agent names remain in examples and guides. Examples and FAQ are extensive, but many are illustrative expected conversations or claimed outcomes rather than evidence of actual behavior. The docs mention disabling skills and selective installation, but give limited coverage of known limitations, operational boundaries, or performance cost. The root MIT license is complete and explicit. The README claims v2.7.0 and provides release notes, but dates differ across files and no complete changelog is included in the packet. README/CLAUDE.md name a maintainer, the FAQ gives contact details, and GitHub issue/discussion channels are mentioned; no response commitment or ongoing maintenance process is shown.
Output examples include file and line references, severity, and suggested fixes; commit messages also have a usable structure. The combination of skills, specialist agents, commands, templates, and workflows may add integration value beyond generic prompting, though some feature claims lack implementation evidence. Multiple components and automatic triggers may add prompt noise, tool use, and maintenance costs. The docs suggest lowering priority or disabling skills but provide no cost measurement or strong evidence of predictable controls.
README claims about quality scores, validation coverage, and production readiness are mostly self-reported. The supplied CI workflow is traceable to YAML/frontmatter and workflow-schema checks, but does not verify the claimed quality rating or feature correctness. Several documents repeat capabilities and installation instructions but conflict on versions, names, and counts, so cross-source corroboration is limited. The evidence includes concrete configuration and CI steps, alongside unsupported claims such as automatic operation, production readiness, and real-world testing; fact and inference are not consistently separated.
- Before installation or use, check the actual component inventory and paths; the supplied docs disagree on skill, agent, and command counts and names.
- Trigger and permission descriptions conflict. Inspect actual configuration and scripts for components that can write files or use Bash, and review proposed changes.
- Quality, production-readiness, and test-coverage claims are largely project assertions; this assessment is based on static material and did not execute validations.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
v2.6.0 v2.6.0 - Quality Improvements & Design Category Transformation
Release notes excerpt
Release Notes - Claude Code Tresor v2.6.0 > Quality Improvement Release: Design Category Transformation & Enhanced Examples > > Release Date: November 15, 2025 > Version: 2.6.0 > Focus: Content quality improvements and consistency > Status: ✅ PRODUCTION READY --- 🎉 What's New Major Quality Improvements 📈 Overall Quality: 7.1/10 → 9.0/10 (+26.8% improvement) 🎨 Design Category Transformation: - Complete restructuring of all 3 design agents - Quality improvement: 4.0/10 → 8.0/10 (+100%) - From worst-performing to best-in-class category …View on GitHub ↗ -
v2.5.0 v2.5.0 - Agent Reorganization & Extension
Release notes excerpt
Release Notes - Claude Code Tresor v2.5.0 > Major Release: Agent Reorganization & Extension > > Release Date: November 15, 2025 > Version: 2.5.0 > Status: ✅ PRODUCTION READY --- 🎉 What's New Major Features 🏗️ Complete Agent Reorganization - 133 agents organized into 10 team-aligned categories - 40+ functional subcategories for precise specialization - Color-coded visual system for easy navigation - Professional directory structure matching real organizations 📦 Extended Agent Library …View on GitHub ↗
FAQ
Does it cost anything to use?
Can I use it directly with Codex or another coding assistant?
How do I start after installation?
./scripts/install.sh, then try /scaffold, /review, or @test-engineer in Claude Code.Which components run automatically?
@agent invocation, and commands start with slash names.How do I update the utilities?
./scripts/update.sh and documents options including --check and --commands-only.Is Claude Code Tresor Toolkit open source?
Is Claude Code Tresor Toolkit safe to use?
Is Claude Code Tresor Toolkit still maintained?
Who makes Claude Code Tresor Toolkit?
What is Claude Code Tresor Toolkit's official website?
Thanks — we use this to improve the page.