Subagents with their own prompt and tools. Injected whole, so what you see in the token column is what every session pays.
Agent Claude Code
Verify that a bug fix actually resolves the defect described in BUG.md and doesn't introduce obvious regressions. Spawned by /fix-bug before PR creation as the blocking gate for bug-fix branches.
owlot/claude-dnd-dungeon-master
Agent Claude Code
Writes the narrative story draft for a session. Reads session logs and character files to produce an immersive, novel-chapter-style story.md file — prose narrative, anchor plan, and memoir entries for all PCs. Stops at the story draft — does not trigger website generation. Use when the session manager signals logs are…
Agent Claude Code
GSD-Lite Protocol — Pair programming with AI agents while maintaining ownership of reasoning and decisions.
Agent Claude Code
This document provides a complete migration plan for converting the existing command-based system (.claude/commands/) to the new intelligent agent-based system (.claude/agents/). The migration preserves all functionality while adding natural language understanding, intelligent coordination, and improved…
Agent Claude Code
Use this agent when UI components, layouts, styles, or design-related code has been written or modified and needs review for Material Design compliance and dark theme best practices. Also use when the user asks for feedback on UI/UX decisions, color choices, spacing, typography, or accessibility in a dark theme…
Agent Claude Code
Use when assembling and optimizing multi-agent teams to execute complex projects that require careful task decomposition, agent capability matching, and workflow coordination.
Agent Claude Code
Execute implementation instructions delegated by agent mode. Use for coding tasks that need file edits, shell commands, and verification.
Agent Claude Code
Active triager — normalizes a raw bug issue into a fix-ready L1 problem or rejects it. First pass on every bug-labeled issue; re-entered on harness kick-back with the failed attempt as evidence. Never fixes, never localizes on first pass.
aws-samples/sample-aidlc-design-reviewer
Agent Claude Code ✓ vendor
Alternatives phase of the AIDLC design review. Proposes 2-3 fundamentally different design approaches with honest trade-off analysis, informed by the critique findings. Returns strict JSON. Invoked by the /design-review skill — not usually run directly.
Agent Claude Code
Specialized subagent for ingesting YouTube transcripts into NexusDB as a single high-fidelity study note while enforcing the YouTube Knowledge Ingestion skill, validation pipeline, and promotion workflow.
eliottvalette/Anthropic-Voodoo-Hackathon
Agent Claude Code
Headless browser tester and root-cause diagnostician for the proto-pipeline-m single-file HTML playables. Runs Playwright on the playable, captures the state trajectory and console errors, cross-references symptoms with the source sketches in 04sketches.json and the 04plan.json contract, and returns a structured…
Agent Claude Code
Unified UI/UX design agent. Implements professional-grade interfaces using Galaxy (3,000+ components), shadcn/ui (accessible React), and UI/UX Pro Max (161 rules, 50+ styles, 161 palettes, 57 font pairings). Covers web and mobile design. Uses Lightpanda for visual verification and nano-banana for image generation.
Agent Claude Code
This directory contains the 12 Claude Code agents that implement the SpecRoute skeleton. They are project-internal - they build SpecRoute itself, not consumer-facing templates that ship as artifacts.
bitia-srl/bitia-vs-code-copilot-examples
Agent Claude Code
Reviews code for vulnerabilities and unsafe defaults.
Agent Claude Code
Expert ArcGIS API for Python GIS Developer. Use when: any ArcGIS task — query, publish, spatial analysis, admin, geometry, geocoding, raster, network, geoenrichment, deep learning. Tries arcgis-mcp/ tools first; falls back to direct arcgis Python API execution when MCP tools don't cover the request. Uses arcgisdocs…
Pixel-Process-UG/superkit-agents
Agent Claude Code
Evaluates subjective quality criteria via LLM-as-judge pattern — scores against defined rubrics and provides structured pass/fail with reasoning for both deterministic and non-deterministic validation.
Agent Claude CodeCursor
Frontend developer. Use when implementing or fixing UI, forms, results, admin pages, or client-side integration per architect and PO specs.
Agent Claude Code
Start codebase analysis. Analyze single entity/action only. Return detailed analysis, recommendations, implementation guidance, usage examples. Prefer dependencylibraries-related-files-search first. Provide all relevant context (files, paths, symbols, imports, etc.). One topic or one context per request → wait for the…
Agent Claude Code
Human-gated code review orchestrator that bootstraps change context, scopes hotspots, picks perspectives and depth, and merges skill-backed perspective findings into one report.
Agent Claude Code
Reviews a diff in a fresh context for correctness, task match and missing tests. Does not edit files.
RaNDoM6913/claude-code-superkit
Agent Claude Code
Review Telegram bot code against SocialApp bot patterns and conventions.
BostonOrange/claude-code-framework
Agent Claude Code
Synthesizes findings from parallel reviewer agents into one consolidated report — dedupes, filters false positives, applies cross-iteration state, classifies the diff into a risk tier, and decides which sub-agents to spawn.
Agent Claude Code
Enumerates the honest alternatives for one open question into one scratch file, invoked with its Short Title, milestone directory, and scratch directory.
Agent Claude Code
System-prompt body for the journey-simulation skill's Phase 3 runner, a headless process the skill starts. Reach it only through that skill — never through the Agent tool, never for general work. Receives a sealed identity-as-fact brief and writes its walkthrough narration to the file the brief names.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: