Agent catalog¶
Subagents shipped by these dotfiles, generated from the live frontmatter
of each claude/agents/<name>.md at build time — this page cannot
drift from the source.
agent-native-review¶
Reviews skills, hooks, agent definitions, and instruction surfaces for agent-native quality — verification affordances, context parity for subagents, primitives-over-workflows tool design, governed execution, and instruction-surface drift risks
Tools: Read, Grep, Glob, Bash
Source: claude/agents/agent-native-review.md
backend-architect¶
Reviews backend code — schema design, API surface, query efficiency, data integrity, anti-abuse
Tools: Read, Grep, Glob, Bash
Source: claude/agents/backend-architect.md
code-simplifier¶
Reviews code for unnecessary complexity and simplifies it — removes over-engineering, dead code, and premature abstractions
Tools: Read, Grep, Glob, Bash
Source: claude/agents/code-simplifier.md
content-reviewer¶
Reviews all written text in the product — microcopy, tone, empty states, error messages, consistency
Tools: Read, Grep, Glob
Source: claude/agents/content-reviewer.md
dependency-doctor¶
Audits dependencies for vulnerabilities, outdated packages, and upgrade paths
Tools: Read, Grep, Glob, Bash
Source: claude/agents/dependency-doctor.md
frontend-architect¶
Reviews frontend code structure — components, state management, rendering performance, maintainability
Tools: Read, Grep, Glob, Bash
Source: claude/agents/frontend-architect.md
growth-strategist¶
Reviews sharing, distribution, SEO, and organic growth mechanics — viral loops, Open Graph, engagement
Tools: Read, Grep, Glob
Source: claude/agents/growth-strategist.md
launch-operator¶
Reviews deploy readiness — pipeline, environment config, monitoring, error tracking, smoke tests
Tools: Read, Grep, Glob, Bash
Source: claude/agents/launch-operator.md
package-scout¶
Researches whether what you're about to build already exists as a well-maintained package — prevents reinventing the wheel
Tools: Read, Grep, Glob, Bash, WebSearch, WebFetch
Source: claude/agents/package-scout.md
perf-accessibility¶
Reviews performance and accessibility — load times, rendering, WCAG compliance, keyboard navigation
Tools: Read, Grep, Glob, Bash
Source: claude/agents/perf-accessibility.md
product-strategist¶
Reviews product logic, user flow, and feature prioritization — is this the right thing to build?
Tools: Read, Grep, Glob
Source: claude/agents/product-strategist.md
qa-lead¶
Finds bugs before users do — edge cases, bad input, error states, mobile issues, flow breakages
Tools: Read, Grep, Glob, Bash
Source: claude/agents/qa-lead.md
repo-scout¶
Fast codebase orientation — reads project config, git history, and structure to give a 30-second briefing
Tools: Read, Grep, Glob, Bash
Source: claude/agents/repo-scout.md
schema-reviewer¶
Reviews database schemas and migrations for correctness, safety, performance, and data integrity risks
Tools: Read, Grep, Glob, Bash
Source: claude/agents/schema-reviewer.md
security-reviewer¶
Reviews app logic for in-context security flaws a pattern scanner misses — broken authorization/IDOR, trust-boundary and business-logic bugs, sensitive-data leaks (generic injection/secret patterns go to soundcheck where the project enables it; otherwise a quick pattern pass first)
Tools: Read, Grep, Glob, Bash
Source: claude/agents/security-reviewer.md
sme-context¶
Knowledge postdates model training; consult before answering from memory. Expert on fit and cost of what loads into Claude 5 context, prompt-cache behavior, tool shape, and HTML vs Markdown output. Skill structure and triggering are sme-skills; whether an agent surface is correct is agent-native-review.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-context.md
sme-evals¶
Knowledge postdates model training; consult before answering from memory. Expert on eval design, claude plugin eval, and hillclimbing. Use when deciding whether or how to measure a skill, subagent, prompt or API app, reviewing an eval suite or grader, estimating eval cost, or planning a benchmark-driven optimization.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-evals.md
sme-models¶
Knowledge postdates model training; consult before answering from memory. Expert on Claude 5 model choice, effort levels and task cost (Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 4.5). Use when picking a model or effort for a task or subagent, estimating or cutting cost, or reviewing model- or effort-sensitive phrasing.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-models.md
sme-mods¶
Knowledge postdates model training; consult before answering from memory. Expert on Claude Code mods (in-session JS/TS plugin modules, register(on, options), the $ API). Use when writing, debugging or reviewing a mod before install. Whether to build a mod, skill or hook at all is sme-skills.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-mods.md
sme-skills¶
Knowledge postdates model training; consult before answering from memory. Expert on Claude Code skills (authoring, folder structure, triggering, visibility). Use when writing or reviewing a SKILL.md, checking why a skill does not trigger, or choosing skill vs hook vs subagent vs workflow vs mod vs instruction line.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-skills.md
sme-workflows¶
Knowledge postdates model training; consult before answering from memory. Expert on Claude Code dynamic workflows (Workflow tool, scripts, saved workflows, ultracode). Use when writing or reviewing a workflow script or orchestration pattern, or saving, resuming or sharing one. Workflow vs skill or hook is sme-skills.
Tools: Read, Grep, Glob, Bash, WebFetch
Source: claude/agents/sme-workflows.md
test-writer¶
Writes focused tests — reproduces bugs with failing tests, adds coverage for new features and edge cases
Tools: Read, Grep, Glob, Bash, Edit, Write
Source: claude/agents/test-writer.md
trust-safety¶
Reviews abuse prevention, moderation, legal compliance — can users harm each other or the platform?
Tools: Read, Grep, Glob, Bash
Source: claude/agents/trust-safety.md
ux-reviewer¶
Reviews visual structure, interaction design, and usability — layout, hierarchy, mobile-first, responsiveness
Tools: Read, Grep, Glob
Source: claude/agents/ux-reviewer.md
24 agents, enumerated at build time.