Subagents de Claude Code · página 9

Specialized agent for verifying software projects and fixing quality failures within the current task scope. Use proactively after code changes or for quality, test, build, lint, format, correctness, or fix requests.

Collects compact scope and cost evidence for requirement confirmation while the user and orchestrator retain requirements, Structural Scale, and document-routing decisions. Use when new requirements, scope, or implementation extent must be confirmed.

Selects this project's smallest task-execution skill set and task-specific safeguards for standalone task and diagnosis workflows. Use when recipe-task or recipe-diagnose needs repository rules for execution.

Discovers functional scope from existing codebase for reverse documentation. Identifies targets through multi-source discovery combining user-value and technical perspectives. Use when "reverse engineering/existing code analysis/scope discovery" is mentioned.

Reviews completed implementation against governing security requirements and the reachable trust model. Use after implementation or when security review/security check/vulnerability check is requested. Returns only must-fix findings with the smallest sufficient corrections.

Derives multiple solutions for verified causes and analyzes tradeoffs. Use when root cause verification has concluded, or when "solution/how to fix/fix method/remedy" is mentioned. Focuses on solutions from given conclusions without investigation.

Converts an approved Work Plan into the fewest executable implementation task files. Use when work plans are approved and task materialization is needed.

Executes React implementation completely self-contained from an explicit prompt or frontend task file. Use when frontend task files exist, or when "frontend implementation/React implementation/component creation" is mentioned. Asks no questions, executes consistently from investigation to implementation.

Executes implementation completely self-contained from an explicit prompt or task file. Use when task files exist in docs/plans/tasks/, or when "execute task/implement task/start implementation" is mentioned. Asks no questions, executes consistently from investigation to implementation.

Creates a scoped frontend ADR batch or one Design Doc from confirmed UI requirements and decision-relevant repository evidence. Use when frontend technical choices or implementation design need an approved artifact.

Creates a scoped ADR batch or one backend/general Design Doc from confirmed requirements and decision-relevant repository evidence. Use when technical choices or implementation design need an approved artifact.

Gathers decision-relevant UI facts from recorded external resources and the existing codebase. Use when frontend design needs compact evidence before UI Spec or Design Doc creation.

Creates UI Specifications from confirmed requirements and optional prototype code. Use when frontend UI design is needed, or when "UI spec/screen design/component decomposition/UI specification" is mentioned.

Critically evaluates investigation results, checks path coverage, and validates failure points using Devil's Advocate method. Use when investigation has completed, or when "verify/validate/double-check/confirm findings" is mentioned. Focuses on verification and conclusion derivation.

Creates implementation-focused work plans from approved Design Docs. Use when Design Doc is complete and implementation planning is needed, or when "work plan/implementation plan/task planning" is mentioned.

|

|

|

|

>-

>-

Use this agent when you need comprehensive full-stack engineering expertise, particularly for server-side development, system architecture, DevOps, and infrastructure tasks. Examples include: (1) Designing and implementing scalable server architectures using languages like Node.js, Python, Go, Java, or Ruby; (2) Setting up CI/CD pipelines, Docker containers, Kubernetes orchestrations, or cloud infrastructure on AWS/Azure/GCP; (3) Optimizing database performance, implementing caching strategies, or designing API systems; (4) Troubleshooting production issues, analyzing logs, or performing system monitoring and alerting setup; (5) Implementing security best practices, authentication systems, or compliance requirements; (6) Conducting code reviews for backend services, infrastructure as code, or deployment configurations. This agent should be consulted proactively when architectural decisions, scalability concerns, or operational excellence considerations are needed for any server-side or infrastructure component.

Use this agent when you need Pencil-based design workflows: atomic designer actions, MCP tools, and spec-driven UI plans. Examples include: (1) Turning PRD or user requirements into action-level PENCIL_PLAN via pencil-ui-design-spec-generator; (2) Executing Pencil MCP tools in sequence (open_document, batch_design, set_variables, get_screenshot, get_style_guide, etc.) to simulate designer operations; (3) Initializing design systems (Layui, Ant Design, Bootstrap, Element Plus, uView, uView Pro, Vant, uCharts, ECharts) with pencil-ui-design-system-* skills; (4) Creating new pencil-ui-design-system-* skills with pencil-skill-creator. This agent should be used when the task involves Pencil, Figma-like design automation, or action-level design plans rather than high-level text-to-UI.

Read-only code review agent. Catches type issues, pattern violations, missing tests. Cannot modify files.

Run tests, analyze failures, suggest fixes. Use after code changes.

>-

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

|

Agent van hanh kenh — thiet lap kenh, brief landing page, email marketing, social listening

Agent san xuat noi dung — viet script, copy, brief creator, lap lich noi dung

Agent chien luoc marketing — lap ke hoach, nghien cuu thi truong, phan tich doi thu, xay dung chien luoc thuong hieu

Agent phan tich hieu suat — doc data, danh gia chien dich, tinh KPI, bao cao

Agent xay dung thuong hieu ca nhan voi AI Avatar — chien luoc, content engine, monetization, community cho founder/coach/creator

Agent visual & production — asset list, brief anh/carousel/video, brand guideline, design review, HTML email/landing

Use this agent when a user reports a concrete bug ("X doesn't work", "crashes on Y", "wrong output for Z", a GitHub issue with steps-to-reproduce). The agent reproduces the bug, locates the root cause, applies a minimal fix, proves the fix works with a regression test, and confirms nothing else broke. Do NOT use for feature requests, refactors, or open-ended "look at this code" asks — those are different profiles.

Use this agent when the user asks an open-ended question about how the ast-index codebase works ("how is incremental update wired", "why don't we tree-sitter Perl", "what's the data flow for --format json", "which commands share scope filtering"). The agent reads the code, traces connections, and returns a structured answer with file:line citations. It does NOT edit code, does NOT run destructive commands, and does NOT speculate — every claim is grounded in a file you can open.

Use this agent to review a set of changes before they ship — staged/unstaged diff, a specific commit range, a PR description, or "the latest commit". The agent reads the diff against project rules in CLAUDE.md and .claude/rules/*, runs the build and test suite, and reports what's good, what's wrong, and what's missing. It does NOT fix anything itself — the point is feedback, not rework.

Use this agent to proofread English text with a focus on formatting and word choice across the project.

Use this agent when you need to create or modify React components following the project's simplicity-first philosophy. This includes building new UI components, refactoring existing components to use the internal UI package, or updating components to follow React 19 patterns.

Use this agent when you need to write or refactor TypeScript code following strict type safety and simplicity principles. This includes creating type definitions, implementing business logic, refactoring JavaScript to TypeScript, or optimizing type inference.

Understands the essence of applications and discovers comprehensive test cases from the perspective of actual user value. Deeply considers how code changes impact user experience and uncovers important test scenarios that are easily overlooked.

Read-only code exploration via Bash + CLI semantic dual-source analysis, with schema-validated structured output.

Compares Decision Digests across role analysis files in a brainstorm session to surface conflicts, gaps, and synergies. Read-only — returns structured text for the orchestrator to apply.

Autonomous executor for non-interactive impeccable commands. Runs audit, polish, harden, layout, typeset, and other automatable design operations without user interaction.

Generates multi-file role analysis for a brainstorm session — analysis.md index + per-feature files + optional findings under {output_dir}/{role}/.

Resident pipeline supervisor agent. Message-driven lifecycle for cross-checkpoint quality observation and health monitoring.

Unified worker agent for team pipelines. Executes role-specific logic loaded from a role_spec file within a built-in task lifecycle (discover, execute, report).

UI design token management and prototype generation — W3C Design Tokens Format, state-based components, WCAG AA validation, responsive layout templates.

Evaluates technical topics, proposals, or decisions across multiple dimensions with evidence-based scoring and recommendations.

Analyzes existing codebase from a specific focus area, spawned in parallel

Collaborative planner working within pre-allocated task ID ranges

Hypothesis-driven debugging with structured evidence logging

Implements single tasks atomically with verification and commit discipline

External research agent using Exa MCP for API details, design patterns, and technology evaluation

Cross-phase integration validation for milestone audits

Test coverage audit with gap detection and test stub generation

Researches implementation approach for a specific roadmap phase

Validates plan quality with up to 3 revision rounds

Creates execution plans with task decomposition, waves, and dependencies

Domain research for project initialization, spawned with different focus angles

Merges multiple researcher outputs into a unified research summary

Multi-dimensional code review agent — analyzes changed files for a single review dimension

Creates project roadmap with phased milestones from research and requirements

Goal-backward verification across three layers (existence, substance, connection)

Deprecated compatibility alias for run-executor

Single-step executor — run next birth packet / run brief(backtrack) + inline skill execution, unnamed nesting for multi-agent orchestration

|

|

Analyzes skill effectiveness data to identify failure patterns and recommend improvements. Use after /skill-monitor flags underperforming skills.