aria-plugin
10CG/aria-plugin
Aria AI-DDD Plugin for Claude Code
Instalação
npx convez-ai 10CG/aria-plugin1 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
README
A tradução para o português deste README ainda está na fila.
English | 中文
Aria Plugin
Version: 1.46.5 | Released: 2026-06-14
AI-DDD methodology plugin for Claude Code — 34 user-facing Skills + 7 internal + 11 Agents + 5 Hooks (incl. default secret-guard)
Prerequisites
- Claude Code installed and authenticated
Installation
# Add marketplace
/plugin marketplace add 10CG/aria-plugin
# Install (Skills + Agents + Hooks included)
/plugin install aria@10CG-aria-plugin
What's Included
Hooks (Auto-triggered)
| Hook | Matcher | Script | Purpose |
|---|---|---|---|
SessionStart | * | session-start-check.sh | Detect interrupted workflows + prompt recovery |
PreToolUse | Write|Edit|NotebookEdit | handoff-location-guard.sh | Rule #9 L1 — block writes to .aria/handoff/*.md |
PreToolUse | Bash | secret-guard.sh | Rule #7 Layer 2 — block raw secret reads (cmd pattern scan), v1.24.0+ |
PreToolUse | Read|Edit|Write|MultiEdit | secret-guard.sh | Rule #7 Layer 2 — block secret-bearing file paths (.env, id_rsa, etc.), v1.24.0+ |
PostToolUse | Bash|Read|Edit|Write|MultiEdit | secret-scan.sh | Rule #7 Layer 2 — REDACT secret-shaped output before reaching LLM context, v1.24.0+ |
Disable Hooks:
# Set environment variable
export ARIA_HOOKS_DISABLED=true
# Or disable the plugin
/plugin disable aria@10CG-aria-plugin
Skills (34 user-facing + 7 internal = 41 total)
Internal skills (7,
user-invocable: false): agent-router, agent-team-audit, arch-common, audit-engine, config-loader, git-remote-helper (v1.15.0 +1), aria-token-telemetry (v1.33.0 +1).
Ten-Step Cycle Core
- state-scanner — Project state scan with intelligent workflow recommendations
- workflow-runner — Lightweight ten-step cycle orchestrator
- phase-a-planner — Phase A planning executor
- phase-b-developer — Phase B development executor
- phase-c-integrator — Phase C integration executor
- phase-d-closer — Phase D closure executor
- spec-drafter — Create OpenSpec proposal.md
- task-planner — Break down OpenSpec into executable tasks
- progress-updater — Update project progress state
Collaborative Thinking
- brainstorm — AI-assisted decision discussion and requirement clarification (problem/requirements/technical modes)
Git Workflow
- commit-msg-generator — Generate Conventional Commits messages
- strategic-commit-orchestrator — Cross-module / batch / milestone commit orchestration
- branch-manager — Branch creation and PR management
- branch-finisher — Branch completion and cleanup
Dev Tools
- subagent-driver — Subagent-Driven Development (SDD) with two-phase code review
- agent-router — Intelligent task-to-agent routing
- tdd-enforcer — Enforce TDD workflow
- requesting-code-review — Two-phase code review (Phase 1: spec compliance → Phase 2: code quality)
Architecture Docs
- arch-common (internal, non-user-invocable) — Architecture tooling shared config
- arch-search — Search architecture documentation
- arch-update — Update architecture documentation
- arch-scaffolder — Generate architecture skeleton from PRD
- api-doc-generator — API documentation generation
Requirements
- requirements-validator — PRD / Story / Architecture validation
- requirements-sync — Story ↔ UPM state sync
- forgejo-sync — Story ↔ Issue sync
- openspec-archive — Archive completed OpenSpec changes (auto-fixes CLI bugs)
Infrastructure
- config-loader (internal, non-user-invocable) — Configuration loading
- git-remote-helper (internal, non-user-invocable) — Git multi-remote parity detection and push verification shared infrastructure (US-012, Layer 3)
- aria-token-telemetry (internal, non-user-invocable) — Context/token telemetry shared data layer (relay cache read + transcript usage parse + window 4-tier resolve; #104, reused by #18 estimator)
Context Awareness (v1.33.0, #104)
- aria-context-monitor — Machine-read current session context occupancy (runtime-truth via statusLine relay) to inform "continue vs pause" decisions
Effort Estimation (v1.34.0, #18)
- ai-native-estimator — Token-axis cycle workload estimation v1 (phase-d auto-capture + forecast/velocity query; Token replaces 4-8h human-hour assumption)
Visualization
- aria-dashboard — Project progress dashboard (UPM/Stories/OpenSpec/Audit/Benchmark)
Environment Diagnosis (v1.24.0)
- aria-doctor — Detect aria-plugin secret-guard hook install state (
check_secret_guard_install5-state schema) + statusLine relay state (check_context_relay3-state + jq, v1.33.0)
Project Adaptation (v1.13.0)
- project-analyzer — Scan project tech stack, frameworks, and work patterns
- agent-gap-analyzer — Compare project needs vs Agent capabilities, identify gaps
- agent-creator — Generate project-specific Agent configs with STCO + capabilities
Feedback & Reporting
- aria-report — Report bugs, feature requests, or questions to the Aria team
Audit System
- audit-engine (internal, non-user-invocable) — Multi-round convergence/challenge audit orchestrator
- agent-team-audit (disabled by default, enable via
.aria/config.json) — Single-round audit executor
Agents (11)
Core Management
- tech-lead — Technical architecture decisions, task planning, cross-team coordination
- context-manager — Multi-agent collaboration, context management
- knowledge-manager — Knowledge base management, documentation sync
- code-reviewer — Two-phase code review (Phase 1: spec compliance + Phase 2: code quality)
Development
- backend-architect — Backend architecture, API design, database schemas
- mobile-developer — React Native / Flutter, offline sync
- qa-engineer — Quality assurance, code review, test strategy
Specialized
- ai-engineer — LLM applications, RAG systems, agent orchestration
- api-documenter — OpenAPI specs, SDK generation
- ui-ux-designer — Interface design, wireframes, design systems
- legal-advisor — Privacy policies, terms of service, GDPR compliance
Usage
Hooks (Auto-triggered)
After installation, hooks fire automatically:
# Session start — detect interrupted workflows
# → checks .aria/workflow-state.json for unfinished work
# PreToolUse Bash — block raw secret reads (v1.24.0+)
# → e.g. `nomad var get ...` without REDACT filter → BLOCKED with helpful stderr
# → bypass: append `# guard:ack: <reason ≥8 non-whitespace chars>` to command
# (audit-logged to ~/.claude/logs/guard-bypass.log)
# PreToolUse Read|Edit|Write|MultiEdit — block secret-bearing file paths (v1.24.0+)
# → e.g. reading .env / id_rsa / .pem / .aws/credentials / .kube/config → BLOCKED
# PostToolUse * — scan output for secret-shaped content, REDACT before LLM (v1.24.0+)
# → warn-only (exit 0 always); replaces secret values in tool_response
# Diagnose install state:
bash ${CLAUDE_PLUGIN_ROOT}/skills/aria-doctor/scripts/check_secret_guard_install.sh
# → JSON state: not_installed / single_plugin / single_local / dual_install / corrupted_settings
Manual Invocation
# Skills
/aria:state-scanner
/aria:spec-drafter
/aria:workflow-runner
/aria:brainstorm
/aria:requesting-code-review
/aria:report bug
# Agents
/aria:tech-lead
/aria:backend-architect
/aria:code-reviewer
/aria:knowledge-manager
Aria 2.0 — Autonomous Runtime
This plugin (aria-plugin) is the interactive layer of Aria — interactive Skills + Agents for use with Claude Code.
For the Aria 2.0 autonomous runtime (10CG Lab internal infrastructure that executes Aria methodology cycles autonomously, currently in development per US-026 M6 milestone), see:
- Aria main repository — methodology + Aria 2.0 PRD + autonomous runtime documentation
- aria-orchestrator — Layer 1 (Hermes + Luxeno-routed GLM models) + Layer 2 (aria-runner + Claude Code + aria-plugin) implementation
aria-plugin does NOT bump to v2.0 when Aria v2.0.0 releases — semantic boundary preserved (plugin = universally available interactive tools; Aria main repo = methodology + 10CG Lab-internal runtime). Plugin users: no action needed. See docs/release-notes-v2.0.0.md §Plugin Compatibility in the Aria main repo for full semantic boundary explanation.
Related Projects
- Aria — Aria main project (methodology research + Aria 2.0 autonomous runtime)
- aria-standards — Aria methodology standards
- aria-orchestrator — Aria 2.0 autonomous runtime (10CG Lab internal)
License
MIT — 10CG Lab