Recommended Free Tools
CLAUDE.md can explain your architecture to Claude Code, but it cannot reliably enforce that architecture by itself. Use it for concise project context; use hooks and permissions to control tool actions, architecture tests to check code dependencies, and sandboxing with least privilege to limit what an agent can reach. Each safeguard addresses a different failure mode, and none certifies the whole system.
What CLAUDE.md can—and cannot—do
Anthropic describes CLAUDE.md as project memory: a briefing file that helps Claude Code understand how a repository works. It can capture build, test, lint, and run commands; coding conventions; major architectural components and how they communicate; important constraints; and known gotchas. That makes it useful onboarding context, not a formal architecture specification. Anthropic’s guidance on memory recommends keeping the file concise, checking generated material for accuracy, and pruning stale instructions.
As an Amazon Associate I earn from qualifying purchases.
A practical guideline is roughly 200 lines or fewer, not a technical limit. Anthropic’s June 2026 guidance also recommends keeping the root file concise, assigning it an owner, and reviewing edits like code. In a larger repository, put directory-specific guidance in path-scoped rules and reserve skills for procedures that should load when needed. Anthropic’s steering guidance explains how these mechanisms serve different purposes.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The key limitation is that a model-facing instruction influences choices; it does not mechanically reject a forbidden import, prevent an unapproved edit, or fail a build when a dependency points the wrong way. If a rule must be reliably enforced, it needs a deterministic control or an executable check.
#1 Best Overall
Match each safeguard to the job
| Mechanism | Best used for | What it does not establish |
|---|---|---|
CLAUDE.md |
Repository orientation, durable facts, conventions, architectural intent, and commands. | That the model will always follow the guidance or that code obeys a boundary. |
| Path-scoped rules | Constraints relevant to particular files or directories, such as API or migration rules. | That unrelated work will need or follow those constraints. |
| Skills | Reusable, multi-step procedures such as release or review checklists. | That a procedure is automatically enforced simply because it is documented. |
| Hooks and permissions | Checking, gating, denying, or restricting tool actions. | Comprehensive verification of the architecture of the resulting code. |
| Architecture tests | Executable checks of selected code dependencies and structural constraints. | Limits on the agent’s access to files, secrets, or network destinations. |
| Sandboxing and least privilege | Restricting what commands and subprocesses can access. | Whether code follows the intended module or dependency design. |
Anthropic says a PreToolUse hook can inspect a tool call and deny it, and recommends hooks or permissions when behavior must be reliably enforced. Managed settings can impose organization-wide deterministic controls that individual local settings cannot override. The exact configuration and product behavior can change, so check the current hooks documentation and permissions documentation before implementing a policy.
A hook can block a class of tool call without verifying every architectural property in changed code. Conversely, a dependency test can flag an architectural violation without limiting the agent’s access to a secret or external destination. Treat these controls as complementary, not interchangeable.
Rank #2
Turn important architecture boundaries into tests
Start with a few boundaries whose violation would be expensive or security-relevant. Define which modules may depend on which others, whether cycles are allowed, which layers may call other layers, and whether the code matches a maintained component diagram.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →For Java projects, ArchUnit’s 1.5.1 Java guide documents checks for package and class dependencies, layer access, cycles, and adherence to PlantUML diagrams. It can be integrated with ordinary Java test frameworks. One example rule restricts access to services to controllers and services. A violation then fails an executable test rather than relying on the coding agent to remember a sentence in a briefing file. This is a Java-specific example; teams in other ecosystems need to verify that their chosen tools can inspect the relevant language and build model.
Rank #3
Useful initial rules may include:
- Core or domain logic cannot depend on delivery, database, or framework code.
- Feature modules communicate only through declared public interfaces.
- Cycles across components are forbidden.
- Infrastructure access stays behind designated ports or adapters.
- A deliberately maintained component diagram matches the dependency structure.
These are candidate constraints, not a universal architecture. Choose rules that reflect your own boundaries and can be checked meaningfully. ArchUnit’s architecture-testing overview recommends expressing selected constraints as code, running them with changes, and evolving them as the design changes.
Put the checks in the normal Claude Code workflow
- Document intent: Keep the root
CLAUDE.mdshort. Summarize key modules, permitted dependency direction, relevant commands, and where enforceable rules live; avoid duplicating a large specification in always-loaded context. - Scope local guidance: Put file- or directory-specific constraints in path-scoped rules. Put repeatable multi-step processes, such as a release checklist, in skills.
- Constrain risky actions: Use permissions and hooks for tool actions that must be denied, gated, or checked. Use managed settings when deterministic organization-wide policy must not be locally overridden.
- Test architecture: Encode mechanically checkable boundaries and run the tests through the usual test runner and, where appropriate, CI. Make failures identify the offending dependency or boundary.
- Limit blast radius: Apply filesystem and network restrictions and grant only the access the task needs.
- Review code and guardrails: Inspect changes to both the application and the rules protecting it. Update exceptions and tests deliberately as the architecture evolves.
A passing architecture test means the constraints it checks passed. It does not prove that every desired quality attribute or design decision is correct.
Rank #4
Why security controls still matter
Instructions can be challenged by untrusted content an agent reads. Anthropic describes indirect prompt injection through sources such as email, web pages, uploaded-image OCR, and tool output. Its guidance recommends distinguishing untrusted content from system instructions, limiting access to sensitive data and actions, using sandboxed tools and narrow permissions, and testing workflows with deliberate injection attempts. See Anthropic’s prompt-injection guidance.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsSandboxing addresses access and blast radius, not architecture compliance. Anthropic’s 2025 sandboxing article reported an 84% reduction in permission prompts in its internal usage; that is a company-reported prompt-friction result, not a measure of architecture compliance or prompt-injection prevention. The sandboxing article describes filesystem isolation and network controls for Claude Code’s Bash execution.
Separately, Anthropic’s 2026 auto-mode article reported that Claude Code users approve 93% of permission prompts. This is a vendor-reported figure in a different context; it does not show that approvals are safe or that architecture is protected. The auto-mode article should not be conflated with the sandboxing result.
What a robust setup can actually promise
These measures make selected rules observable and enforceable. They do not guarantee architectural correctness: teams must choose the right boundaries, review exceptional changes, and maintain checks as the system evolves. The reviewed material provides no independent statistic quantifying how much CLAUDE.md alone protects architecture, or a controlled comparison of architecture-violation rates with and without executable tests. The useful standard is therefore not a promised percentage; it is whether each important rule has a clear owner, an appropriate enforcement mechanism, and a failure signal in the workflow where changes are reviewed.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
