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
Fable 5 planner for the /fable skill. Invoked once by /fable with a plan brief — intent, task verbatim, session constraints, file map, pasted load-bearing code, and 1–3 open questions — and returns the plan: numbered steps with risks, checkpoints, assumptions. Resumed at task end for the warm diff-review…
Agent Claude Code
Database and migration specialist. Use for schema design, migrations, query optimization, indexing, materialized views.
Agent Claude Code
Gate-injected, diff-driven documentation reconciler. Dispatched INSIDE an implementation gate (post-implement, after the build/lint/test gate is green and the change is proven reachable, before the review leg) with the applied diff for one work-unit / wave; it triages doc-impact and INCREMENTALLY updates ONLY the docs…
Agent Claude Code
Use this agent when the user wants to save work, sync repos, undo changes, check history, or manage branches. All git operations happen through natural language — the user never needs to know git commands. Examples: Context: User finished working and wants to save. user: "Save my work" assistant: "I'll use…
Agent Claude Code
On-demand canary for how a team works. Measures a window of work against what it delivered (cycle time, token cost and agent time per merged PR or other unit), flags health ratios that look wrong (waiting vs working, test and build time, re-runs, cache writes and avoidable cache misses, context growth), checks whether…
Agent Claude Code
Your co-founder who doesn't exist yet. Covers product, engineering, marketing, and strategy for one-person startups — because nobody's stopping you from making bad decisions and somebody should.
Agent Claude Code
Fast read-only search agent for locating code. Use it to find files by pattern (eg. "src/components//.tsx"), grep for symbols or keywords (eg. "API endpoints"), or answer "where is X defined / which files reference Y." Do NOT use it for code review, design-doc auditing, cross-file consistency checks, or open-ended…
Agent Claude Code
Use when an EM needs to scan/triage across multiple repos, services, or large result sets in parallel — e.g. "which of these 5 repos reference this deprecated API", "find all TODOs related to X across the org's services". Read-only, keeps the dispatching session's context clean.
ryantlee25-droid/spectrum-protocol
Agent Claude Code
Product research and problem identification agent. Validates that a problem is real, significant, and worth solving before solution exploration. Produces a structured Problem Brief grounded in evidence-based discovery frameworks. user: "what problems should we focus on for our dev tools product?" assistant: uses…
Agent Claude Code
Reviews code for bugs, logic errors, security vulnerabilities, code quality issues, and adherence to project conventions, using confidence-based filtering to report only high-priority issues that truly matter.
Agent Claude Code
Use this agent when you need to analyze code structure, identify patterns, or understand codebase architecture. Examples.
Agent Claude Code
An agent that expands a chapter outline into a complete serial-fiction chapter while following the project's style and character files.
brightdata/brightdata-cursor-plugin
Agent Claude Code
Autonomous research agent using Bright Data's full web intelligence stack. Searches, scrapes, and structures web data to answer complex questions with fully cited sources. Use for deep, multi-source research tasks.
Agent Claude Code
Write or update inline docs, README sections, and API references after a code change. Use when code has been added/modified and the docs need to catch up, or when a module is undocumented.
Agent Claude Code
You are the foundational analysis agent. You perform codebase exploration, analyze dependencies, read documentation, research technologies, and generate test scenarios. Your raw analysis outputs feed into the T4 Lead Analyst for consolidation. You never write application code.
Agent Claude Code
Decomposes a session transcript into typed topic-units for the recall index. Two-phase (exhaustive inventory, then render one unit per thread).
Agent Claude Code
Review staged and unstaged working changes for high-impact hygiene and changeability issues. Write findings to assigned review artifact.
zingspark/create-harness-vibe-coding
Agent Claude Code
Use when a repeated failure, user correction, or WF recovery loop triggers memory writing; also use during WF closeout for consolidation. Writes to Harness/memory/, Harness/MEMORY.md, and cross-project global memory.
Agent Claude Code
Use this agent when you need expert-level React/JavaScript/TypeScript development guidance, microfrontend architecture reviews, performance optimization, or code quality improvements. This agent specializes in React best practices, MFE patterns, Redux/state management, testing strategies, and production-ready…
sethyuan/orcanote-agent-skills
Agent Claude Code
Assist with managing and querying notes in Orca Note.
Agent Claude Code
Independent report-only source-comment hunter. Classifies narration and alibis; keeps the innocent list; never edits files.
Agent Claude Code
Background agent that analyzes session observations to detect patterns and create instincts. Uses Haiku for cost-efficiency. v2.1 adds project-scoped instincts.
Agent Claude Code
Use this subagent to write or update test case documents under docs/Test-Case/ / and the matching automation scripts inside whatever test directory already exists under code/. Two modes — audit-driven (decision = add / skip / rewrite): each TC is derived from the PRD's business process + key page functions, assigned a…
Agent Claude Code
Research & development agent. Conducts multi-source technical research, evaluates options, and produces decision records. Use for investigations, technology evaluations, and pre-implementation research.
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: