Instructions drift
Long, vague project files consume context while leaving the important constraints ambiguous.
Claude Code setup kit · July 2026 edition
Stop rebuilding your Claude Code setup in every repo.
A complete, adaptable starting point for clearer instructions, bounded permissions, repeatable workflows, and less babysitting—delivered as plain files your team can inspect and change.
One-time purchase · Instant ZIP after payment · 14-day refund · Free audit runs locally with no upload
3CLAUDE.md templates
7guardrail & workflow hooks
4permission profiles
6 + 5skills + subagents
3GitHub Actions workflows
8playbook chapters
The gap after installation
Claude Code is powerful. A blank configuration is still a blank configuration.
Permissions become guesswork
Teams bounce between repeated prompts and overly broad autonomy without a shared starting policy.
Good habits stay manual
Verification, secret checks, formatting, and PR readiness are easy to forget when they are only advice.
Free local repository check
Use evidence from a real repo.
Select a project folder for an instant readiness snapshot. The scan runs in your browser: no filenames or contents are uploaded.
Audit a repository freeWhat is included
One internally consistent setup, not a pile of snippets.
The full archive contains 45 files and roughly 2,600 lines of playbook and configuration material.
Project instructions
Templates for Next.js/TypeScript, Python services, and monorepos, plus guidance for keeping them lean.
Hooks
Selected destructive-command blocking, protected paths, staged-secret patterns, formatting, test-on-stop, notifications, and compact session context.
Permission profiles
Cautious, standard, autonomous, and CI starting points so the level of agency matches the environment.
Skills & subagents
Verification, feature delivery, flaky tests, PR readiness, dependency updates, incidents, review, security, migrations, tests, and docs.
CI & MCP
Three Claude Code GitHub Action workflows and a bounded MCP starter with credential and context-cost guidance.
Eight-chapter playbook
The reasoning behind every layer, from CLAUDE.md design and permissions to rollout, context cost, and automation.
A guided implementation path
From download to a checked-in starting point.
- ChooseStart from the template and permission profile closest to your stack and risk level.
- AdaptReplace the clearly marked project commands, paths, and policies with your real values.
- VerifyRun the documented trigger checks before trusting a hook or permission boundary.
- CommitKeep the resulting setup with the repository so it travels with the team.
Honest boundary
Built to be adapted, not blindly copied.
Harden My Repo provides configuration evidence and operating guidance, not a security product. It cannot know your architecture, commands, or risk tolerance. Every delivered file is readable and editable, and version-sensitive behavior links to official documentation.
- No subscription or proprietary runtime
- No telemetry in the kit
- Independent; not affiliated with or endorsed by Anthropic
- Verified on Linux in this sprint; shell components target macOS/Linux conventions
Inspect the quality first
The free Lite kit is useful on its own.
It includes a real Next.js/TypeScript CLAUDE.md template, three hooks, one permission profile, and two workflow skills. No email gate.
Choose by the job to be done
Start free. Pay only when the broader setup saves time.
Free
Inspect one starting stack
Use this when you want a working example before deciding whether the full library is useful.
- Browser readiness snapshot
- 1 CLAUDE.md template
- 3 hooks, 1 profile, 2 skills
$9 once
Configure repositories yourself
Use this when you need adaptable starting points across stacks, risk levels, and recurring workflows.
- 3 templates and 7 hooks
- 4 permission profiles
- 6 skills, 5 subagents, CI, MCP
- Eight-chapter implementation playbook
$49 once
Generate repository-specific evidence
Use this when a generic checklist is not enough and you want prioritized local findings plus drafts.
- Local analyzer; no upload
- HTML, Markdown, and JSON
- Coverage and skipped-file ledger
- Remediation drafts + complete $9 kit
Local analyzer + complete kit
Know what your repository is missing before granting more autonomy.
The Local Repo Report runs on your machine, reads bounded configuration evidence, and produces a prioritized HTML, Markdown, and JSON report with inert remediation drafts. Your source is not uploaded and project code is not executed.
One-time purchase · Unlimited internal repositories for one individual or organization · Includes the complete $9 kit · 14-day refund
Local onlyNo network requests, repository upload, or project-code execution.
Evidence firstVersioned checks, bounded scores, coverage, path and line evidence, and explicit unknowns.
Useful outputsSelf-contained report, structured JSON, and review-required configuration drafts.
Four profilesCautious, standard, autonomous, and CI expectations.
Questions
Know what you are buying.
Is it plug-and-play?
It is a working starting point, not a universal configuration. Replace the marked commands, paths, and policies with your project’s actual values.
Do I need Claude Code already?
The $9 setup kit configures Claude Code and does not include Claude Code, API usage, or an Anthropic subscription. The repository audit can also assess Codex-oriented evidence.
Why pay when examples are free?
You can assemble a setup from documentation and scattered snippets. The paid value is an internally consistent structure plus the reasoning and verification path that connect the pieces.
What does the license allow?
Use and adapt the files across projects owned by one individual or organization. Redistribution, resale, and publishing the paid archive are not allowed.
How does the refund work?
Try the kit in a real repository. Within 14 days, submit your receipt email and order reference on the support page for a refund to the original payment method.
Launch edition
Give Claude Code a better starting point.
Complete bundle, instant delivery, plain files you control.