core-claude#

Claude Code hooks and skills for development, published by ByteCode Solutions.

License Pipeline Status Docs Status

Documentation Contents#

Index:

Reference:

What’s included#

Hooks#

  • ``.env`` file protection: a PreToolUse hook blocks any Read, Grep, or Bash tool call that targets a .env file at runtime.

  • Declarative deny rules: a .claude/settings.json ships with matching permissions.deny rules that enforce the same restrictions before any tool is invoked. Every project that installs this plugin must apply these rules to its own .claude/settings.json, the hook alone is not sufficient.

  • ``/develop`` suggestion: a UserPromptSubmit hook detects development-related prompts and injects a recommendation to use /develop before Claude responds, a nudge, not a forced invocation.

Skills#

  • Create Python project: scaffold a new Python project from the ByteCode Solutions blueprint template via /create-python-project [project-name].

  • Run linters: run all linters and type checkers (ty, ruff, mypy, pyright, pylint) on a Python package sequentially via /run-linters <package>. Stops on failure and asks the user whether to fix before continuing.

  • Run tests: run the project test suite via /run-tests. Uses core-tests / manager.py when available, falls back to pytest. Stops on failure, creates one task per failing test, and iterates through fixes one by one.

  • Run coverage: run test coverage via /run-coverage. Uses core-tests / manager.py when available, falls back to pytest --cov. Reports total coverage % and any module below 100 %; offers to create one task per under-covered file to reach 100 %. On test failure delegates to /run-tests.

  • Run security: run security checks via /run-security <package>. Runs bandit for static code analysis and pip-audit for dependency CVE scanning, one at a time. Stops on findings and asks the user whether to fix before continuing.

  • Develop: orchestrate a full development task via /develop [task description]. Enforces TDD, SOLID, DRY, KISS, YAGNI, and LoD throughout. Enters plan mode, reads the codebase, divides the problem, generates Python interface stubs, and writes PLAN.md — stopping to ask the user whenever multiple design approaches are viable. Spawns one agent per component (parallel when independent, sequential when dependent), each following TDD with 100 % unit coverage and /run-linters. After all agents complete, writes integration tests, runs /run-linters, /run-security, /security-review, and /review; commits using Conventional Commits; generates DIAGRAM.md if multi-component; and optionally opens a PR.

Installation#

claude plugin marketplace add https://gitlab.com/bytecode-solutions/core/core-claude.git
claude plugin install core-claude

To update to the latest version:

claude plugin update core-claude@bytecode-solutions

To uninstall:

claude plugin uninstall core-claude@bytecode-solutions

After installing, add the declarative deny rules to each project’s .claude/settings.json:

{
  "permissions": {
    "deny": [
      "Read(**/.env*)",
      "Grep(**/.env*)",
      "Bash(cat */.env*)",
      "Bash(cat .env*)"
    ]
  }
}

Contributing#

Contributions are welcome! Please:

  1. Fork the repository.

  2. Create a feature branch.

  3. Validate the plugin: claude plugin validate .

  4. Submit a pull request.

License#

This project is licensed under the MIT License. See the LICENSE file for details.

Support#

For questions or support, please open an issue on GitLab or contact the maintainers.

Authors#