convez.ai
MCP ServerDevOps & InfraestruturaAvançado
NDDev-it-com

rldyour-claudecode

NDDev-it-com/rldyour-claudecode

Catálogo controlado de plugins, MCP servers e workflows para Claude Code

Ver no GitHub
BomPluginScore · de 0 a 100

Configuração CLI integrada para Claude Code que centraliza plugins, MCP servers, LSP, memória Serena, workflows de segurança, navegação e design, além de agents revisores. Oferece catálogo controlado e versionado de ferramentas de produtividade para desenvolvimento e automação de fluxos de trabalho.

Instalação

$npx convez-ai NDDev-it-com/rldyour-claudecode

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
87%
Último commit
há 2 dias
Versão
1.8.14
Issues abertas
3
Contribuidores
2

Casos de uso

  • Integração centralizada de MCP servers e LSP com Claude Code
  • Automação de workflows de segurança e revisão de código
  • Gerenciamento de skills, comandos slash e agents dedicados
  • Orquestração de ferramentas de browser, design e exploração

Veredito

Solução robusta de orquestração para quem precisa centralizar múltiplas ferramentas de IA no Claude Code com controle rigoroso de versionamento e segurança. Adequado para times que já dominam MCP/LSP e buscam automação sofisticada além do básico.

README

A tradução para o português deste README ainda está na fila.

rldyour-claudecode

rldyour-claudecode is a rldyour AI CLI configuration for Claude Code: plugin marketplace, MCP/LSP, Serena memory, security review, browser/design workflows, and reviewer agents. It is authored by Danil Silantyev (github:rldyourmnd), CEO NDDev. It is not a generic preset, not an automatic configuration takeover, and not a bundle of unrelated third-party opinions. It is a controlled catalog for the owner's own plugins, MCP servers, skills, subagents, slash commands, hooks, rules, and workflows.

Main principle: nothing is treated as enabled or correct unless the owner explicitly decides it.

License

This project and all first-party marketplace plugins are licensed under the GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). See LICENSE for the full text.

Control Model

  • The active marketplace contains only plugins that are actually created and ready to install.
  • Each plugin must have a clear responsibility boundary (one domain per plugin; no catch-all plugins).
  • Each tool or workflow must describe its purpose, access model, risks, and usage rules.
  • Repository documentation is written in English. User-facing communication with the owner is Russian unless explicitly requested otherwise.
  • Technical identifiers stay stable, ASCII, and kebab-case where applicable.
  • Every callable rldyour skill includes compact Russian and English trigger phrases in the SKILL.md frontmatter description. Claude Code uses descriptions as the primary routing surface; details belong in the skill body or references.
  • Secrets, tokens, cookies, and private keys are never stored in this repository.

Active Catalog

9 first-party plugins (38 skills, 11 slash commands, 8 subagents, 10 hook scripts in 2 manifests, 51 scripts total, 16 references). CI: 11 workflows + 1 dependabot config. Serena memories: 24.

PluginVersionSkillsCommandsAgentsHooksScriptsReferencesMCPLSP
rldyour-mcps1.1.4500000010
rldyour-explore1.1.4521100000
rldyour-serena-mcp1.1.4520143000
rldyour-security1.1.4521000000
rldyour-browser1.1.4560000000
rldyour-design1.1.4551000000
rldyour-lsps1.1.4540002301
rldyour-flow1.1.45107668700
rldyour-rules1.1.4571000600

Marketplace boundary VERSION and per-plugin versions are listed in the auto-generated inventory above (python3 scripts/generate_inventory.py to refresh).

Unified cross-tool behavior is declared in config/rldyour-contract.json and projected to docs/contract-matrix.md. scripts/validate_contract.py proves the local Claude Code adapter against real plugin, skill, command, agent, hook, and CI files; scripts/generate_contract_matrix.py --check keeps the human-readable matrix generated.

Briefly:

  • rldyour-mcps - single-owner MCP transport (11 pinned servers: Serena, Sequential Thinking, Chrome DevTools, Context7, DeepWiki, Grep, shadcn, Dart/Flutter, Figma, OpenAI Docs, GitHub).
  • rldyour-serena-mcp - Serena-first semantic code workflow, numbered fact-only .serena memory sync via flow-memory-sync subagent, lifecycle hooks (UserPromptSubmit, PreToolUse:Bash, PostToolUse:Bash, Stop).
  • rldyour-flow - autonomous SDLC orchestration with seven slash commands (ry-init, ry-start, ry-newp, ry-review, ry-deploy, ry-sync, ry-repair), six reviewer subagents (architecture/quality/consistency/integration/verification/security tracks), advisory SessionStart, PreToolUse:Bash, PostToolUse:Bash, and Stop hooks, scoped context packs, instruction docs sync, and post-task synchronization.
  • rldyour-explore - deep multi-source research via ry-explore subagent (model: opus[1m], effort: max) and tech/web research skills routing through Context7, DeepWiki, Grep, and authoritative web sources.
  • rldyour-security - non-blocking OWASP Top 10 2025 secure-implementation guidance plus the ry-sec-review defensive review skill.
  • rldyour-browser - provider-routed browser workflows for Webwright, Playwright CLI, and Chrome DevTools MCP.
  • rldyour-design - Figma → code, centralized token-based design system, strict Feature-Sliced Design frontend architecture, shadcn/ui, ReactBits, and browser-validation workflows.
  • rldyour-lsps - language-server routing, health checks, brew-first setup profiles, and Serena LSP integration guidance.
  • rldyour-rules - quality-first engineering rules, architecture boundaries, implementation discipline, dependency compatibility (SLSA Level 2, SBOM, lockfile discipline), verification gates, project-instruction policy, MADR 4.0.0 ADR policy, and ry-rules-review.

What Claude Code Reads

Claude Code reads:

  • .claude-plugin/marketplace.json - active installable plugin catalog using per-entry relative sources (source: "./plugins/<name>"); this repo intentionally does not use metadata.pluginRoot.
  • plugins/<plugin>/.claude-plugin/plugin.json - plugin manifest. Each manifest declares dependencies as an array; rldyour-mcps is the base layer with no dependencies, all other plugins depend on it (and rldyour-flow additionally depends on rldyour-serena-mcp).
  • Manifest-linked files: skills/<skill>/SKILL.md, agents/<agent>.md, commands/<name>.md, hooks/hooks.json, .mcp.json, references, scripts.

This README is for owner review and repository orientation. It explains the control model and active plugin catalog but does not enable tools by itself.

Adding A Plugin

Before creating a new plugin, define:

  • exact plugin name (kebab-case, namespaced under rldyour-);
  • purpose and boundary (single domain, no catch-all);
  • files it will contain;
  • whether it provides MCP servers, skills, hooks, subagents, slash commands, or documentation only;
  • what Claude Code can do through it;
  • what actions require confirmation;
  • what data must never be sent outside the machine or repository.

Only then create plugins/<name>/ with the canonical layout (.claude-plugin/plugin.json, skills/, agents/, commands/, hooks/, references/, scripts/ as needed) and add it to .claude-plugin/marketplace.json.

Local Installation

claude plugin marketplace add /path/to/rldyour-claudecode
claude plugin install rldyour-mcps@rldyour-claudecode
claude plugin install rldyour-serena-mcp@rldyour-claudecode
claude plugin install rldyour-flow@rldyour-claudecode
# ...etc for each enabled plugin

After changing marketplace.json, a plugin manifest, hooks, skills, agents, or .mcp.json, restart Claude Code so the runtime reloads plugin definitions:

claude plugin validate .                     # validate marketplace
for p in plugins/*/; do
  claude plugin validate "$p"                # validate each plugin
done
scripts/validate_marketplace.sh              # full marketplace harness

plugins/rldyour-mcps/.mcp.json is the portable source of truth for MCP server definitions. Local MCP launcher packages are pinned in .mcp.json and additionally documented in config/mcp-runtime-versions.env. Do not use @latest or unpinned uvx --from package specs for local MCP runtime definitions; update versions intentionally and rerun capability smoke tests.

Smoke Tests

The marketplace ships with smoke coverage for MCP runtime startup, MCP capability handshake, hook lifecycle, and the fullrepo synchronization state machine:

scripts/smoke_mcp_runtime.sh                 # MCP servers reachable + pins parsed (~30s)
scripts/smoke_mcp_capabilities.sh            # JSON-RPC initialize + tools/list per server (~2m cold, ~30s warm)
scripts/smoke_mcp_capabilities.sh --server <name>   # narrow to one server
scripts/smoke_mcp_capabilities.sh --skip-uvx        # skip slow uvx cold-starts
scripts/smoke_hooks.sh                       # Serena + flow hook scripts dry-run
scripts/smoke_serena_memory_taxonomy.sh      # analyzer schema/targets + memory freshness edge cases
scripts/smoke_fullrepo_sync.sh               # fullrepo branch lifecycle state machine

Targeted state checks:

plugins/rldyour-flow/scripts/flow_post_task_state.py | python3 -m json.tool
plugins/rldyour-serena-mcp/scripts/serena_memory_state.py | python3 -m json.tool
plugins/rldyour-flow/scripts/instruction_docs_state.py --json | python3 -m json.tool
python3 scripts/validate_instruction_docs.py --require-agent-docs
python3 scripts/validate_plugin_versions.py
python3 scripts/validate_contract.py
python3 scripts/generate_contract_matrix.py --check
python3 scripts/validate_skill_routing.py
python3 scripts/check_mcp_runtime_versions.py
python3 scripts/release_manifest.py

GitHub Actions runs claude plugin validate, JSON/Python/shell syntax checks, and frontmatter verification on every push and pull request via .github/workflows/validate.yml. The scheduled dependency-check.yml workflow monitors pinned MCP runtime versions for upstream drift.

Fullrepo Branch

fullrepo is the portable complete-state branch for agent-only files (e.g. AGENTS.md, .claude/CLAUDE.md, .serena/project.yml, .serena/memories/**). Normal project branches keep product history clean and exclude these files through .git/info/exclude.

This is the default rldyour-owned repository policy. In external or colleague-owned repositories, .rldyour/project-policy.json is the executable source of truth and may disable fullrepo, allow instruction docs on normal branches, and disable branch-cleanup blockers.

python3 plugins/rldyour-flow/scripts/fullrepo_sync.py --bootstrap-init   # first-time setup on a fresh checkout
python3 plugins/rldyour-flow/scripts/fullrepo_sync.py --status-json      # machine-readable sync state
python3 plugins/rldyour-flow/scripts/fullrepo_sync.py --restore          # restore agent-only files from origin/fullrepo
python3 plugins/rldyour-flow/scripts/fullrepo_sync.py --migrate-main     # git rm --cached agent-only files left in the index
python3 plugins/rldyour-flow/scripts/fullrepo_sync.py --publish          # refresh fullrepo with safe --force-with-lease

Customer-facing branch is main; the team works against fullrepo.

Local product repositories that consume this marketplace can install the rldyour Git pre-push guard:

scripts/install_local_git_hooks.sh --dry-run
scripts/install_local_git_hooks.sh --apply

The guard is branch-aware. Product branches keep strict protection against agent-only paths and AI-marker additions. The configured fullrepo branch (fullrepo by default, or RLDYOUR_FULLREPO_BRANCH) allows durable AI context while still blocking definite secrets, runtime markers, browser artifacts, and local env files.

Release, Rollback, And Observability

Marketplace release version lives in VERSION. Per-plugin behavior versions stay in plugins/<plugin>/.claude-plugin/plugin.json. Release notes live in CHANGELOG.md (Keep-a-Changelog format).

python3 scripts/release_manifest.py            # build release metadata bundle
python3 scripts/check_mcp_runtime_versions.py  # detect drift against config/mcp-runtime-versions.env
scripts/collect_diagnostics.sh                 # local ignored diagnostics bundle for failure triage

Reference documents:

  • docs/adr/ - Architecture Decision Records (MADR 4.0.0). Twelve ADRs cover irreversible decisions: fullrepo policy, dual-doc split, bilingual descriptions, reviewer transport, local GitHub MCP, ownership boundaries, MCP pinning, CI baseline, release/tag conventions, macOS egress trust gap, hook freshness invariants for branch-split projects, and the owner full-auto standard. Start with docs/adr/README.md.
  • docs/release-process.md - versioning, CHANGELOG, release evidence, claude plugin tag --push flow.
  • docs/rollback-restore.md - safe restore from previous tags or fullrepo snapshots.
  • docs/dependency-updates.md - pinned MCP runtime update policy.
  • docs/observability.md - diagnostics, CI artifacts, hook lifecycle debugging, failure triage.

Minimum Claude Code Version

v2.1.169 as the pinned CI/runtime baseline and v2.1.146+ as the feature compatibility floor. This covers every Claude Code feature used by the marketplace: opus[1m] for Opus 4.8 extended context (v2.1.154+; [1m] availability remains account/plan-dependent and /model omits it when unavailable), alwaysLoad: true MCP option (v2.1.121+), claude plugin tag --push (v2.1.118+), hook if filter (v2.1.118+), exec-form hook args (v2.1.139+), displayName marketplace entries (v2.1.143+), Stop/SubagentStop background_tasks and session_crons input fields (v2.1.145+), MCP pagination fixes, Auto mode AskUserQuestion behavior needed by decision gates (v2.1.146+), disallowed-tools frontmatter for skills/slash commands, SessionStart reloadSkills output, and the MessageDisplay hook event (v2.1.152), skipLfs marketplace-source option, status line terminal-size env (COLUMNS/LINES), claude agents native command/bundled skill autocomplete (v2.1.153), Opus 4.8 support, dynamic workflows, default streaming tool execution, defaultEnabled: false plugin metadata, piped MCP pending-approval reporting (v2.1.154), the Opus 4.8 thinking-block API hotfix (v2.1.156), current latest package drift through v2.1.169, fallbackModel, --fallback-model, deny rule glob support including "*", relayed SendMessage permission hardening, explicit thinking disable controls, managed allowedMcpServers / deniedMcpServers ${VAR} matching, maxSkillDescriptionChars and skillListingBudgetFraction user settings (v2.1.105+), skillOverrides (v2.1.129+; plugin skills exempt), experimental.{themes,monitors} wrapper (v2.1.129+).

Claude Code v2.1.163 managed runtime bounds, /plugin list, Stop/SubagentStop hookSpecificOutput.additionalContext, skill \$ escaping, and resumed stdio MCP session-id parity are tracked in references/claude-surface-adoption.md. The v2.1.166 target adds runtime/model/permission policy surfaces; the v2.1.168 target is a reliability rollup with no new adapter-owned config surface, and v2.1.169 is a package/runtime rollup adopted from npm latest.

Support And Feedback

This is a personal marketplace - response time is best-effort, no SLA. Feel free to fork and tailor to your own workflow.

Ver o original no GitHub

Bom para

Relacionados