claude-code-config

Introduces what this repository installs and maintains: Claude Code settings, hook scripts, orchestrator rules, specialist agents, plugins, skills, and the bundled CLI.

Explains the orchestrator-versus-specialist model, slash-command execution mode, rule enforcement, task tracking, and the mandatory review loop.

Maps the major directories and shows how agents, rules, scripts, plugins, skills, the Python package, and tests fit together.

Lists runtime prerequisites inferred from the code and hooks, including Python 3.10+, uv, gh, jq, gawk, prek, mcpl, required review plugins, and optional acpx support.

Shows how to install the configuration into a Claude Code home directory, wire up the repo's settings and scripts, install the bundled CLI, and enable the bundled plugins.

Explains the session-start validation flow, how missing tools and critical plugins are detected, and how to confirm the configuration is active.

Shows how to use the included CLAUDE.md template for personal preferences and targeted overrides without changing shared repository files.

Documents permissions, allowed tools, hooks, enabled plugins, environment flags, status-line configuration, and the duplicated allowlist entries that must stay in sync.

Covers the behavior of rule-enforcer.py, git-protection.py, rule-injector.py, session-start-check.sh, and my-notifier.sh, including what each one blocks, warns about, or enriches.

Summarizes the rule files that govern delegation, issue-first workflow, agent routing, MCP discovery, task usage, slash-command handling, parallelism, temp files, and bug reporting.

Profiles the bundled agents, their scope boundaries, notable instructions, and how shared base-agent rules differ from orchestrator rules.

Documents the bundled skills and templates, including browser automation, docsfy-based documentation generation, and the personal Claude configuration template.

A task-oriented guide to how a configured Claude session is expected to operate from request intake through delegation, review, testing, and completion.

Covers local diff review, GitHub PR review, multi-source review handling, pending-review refinement, reply posting, auto-skip behavior, and review result storage.

Explains release validation, commit analysis, version-file detection, version bumping, PR-based release prep, and GitHub release creation.

Shows how the acpx plugin works in exec, fix, and peer modes, including session handling, dirty-worktree safeguards, and parallel multi-agent runs.

Explains how this configuration expects users and agents to discover and use MCP tools through mcpl rather than direct tool guessing.

Documents the project's public interfaces, which are CLI-, slash-command-, hook-, and JSON-based rather than a standalone HTTP API.

Reference for all bundled slash commands, their arguments, allowed tools, and the workflows they trigger across the myk-github, myk-review, and myk-acpx plugins.

Reference for myk-claude-tools reviews commands such as fetch, post, pending-fetch, pending-update, and store.

Reference for PR-oriented commands that fetch diffs, load CLAUDE.md content, and post structured review comments.

Reference for release info, create, detect-versions, and bump-version commands, including their inputs and JSON outputs.

Reference for analytics commands that query the review database, find dismissed patterns, and compare similar comments.

Reference for CodeRabbit rate-limit detection and re-trigger commands used by the GitHub review workflows.

Documents temporary JSON artifacts, hook payload expectations, plugin and marketplace metadata, and the SQLite review database schema.

Explains the Python package setup, dependency groups, local development flow, and repository conventions such as the whitelist-style .gitignore.

Covers pytest, tox, ruff, mypy, flake8, markdownlint, secret scanning, and the behavioral guarantees encoded in the test suite.

Explains the project's automation story, including pre-commit.ci settings, CodeRabbit and Qodo Merge integration, and the fact that no repository-local CI workflow is checked in.

Shows how to add or remove agents, rules, scripts, slash commands, and marketplace entries while keeping whitelist and bug-reporting coverage in sync.

Documents how the repo versions the CLI and plugins, detects version files, and uses release helpers to prepare and publish new versions.

Collects common operational failures such as blocked hooks, missing tools, gh auth problems, CodeRabbit parsing edge cases, acpx session failures, and stale task or temp-file state.

Summarizes the project's MIT license and what it permits for reuse, modification, and redistribution.