figma
25Agent Claude Code
Figma Analyst — Extracts frames from a design tool via MCP, cross-references against requirements, and surfaces design-vs-PRD gaps. Invoked by auditexpert only.
An open-source agentic vibe coding platform that helps you build your own platform (Claude Code, GLM, Codex)
This repository also configures its own agents. See what termstack tells them →
Agent Claude Code
Figma Analyst — Extracts frames from a design tool via MCP, cross-references against requirements, and surfaces design-vs-PRD gaps. Invoked by auditexpert only.
Agent Claude Code
Impact Analyzer — Maps module dependencies and surfaces cross-feature or platform concerns before implementation begins.
Agent Claude Code
Integration Test Writer — Generates DI-wired integration tests with multi-endpoint mock HTTP engines.
Agent Claude Code
Low-Level Design Planner — Produces detailed implementation blueprints for approved tasks. Designs classes, interfaces, and data flows but does NOT write code.
Agent Claude Code
Performance Analyst — Analyze code for performance bottlenecks, optimize critical paths, and provide measurable improvement recommendations. Read-only analysis, hands off to implementers for fixes.
Agent Claude Code
DAG task planner. Decomposes features into atomic, deliverable-focused tasks with pre-mortem analysis, acceptance criteria, and dependency ordering.
Agent Claude Code
PRD Analyst — Parses product requirements documents, extracts structured requirements, and surfaces gaps between PRD, design, and API contracts.
Agent Claude Code
This document defines the mandatory and recommended fields/sections for every PRD processed by the prdanalyst agent. Use this as a checklist to flag missing or underspecified requirements.
Agent Claude Code
Multi-pass codebase and latest-guidance researcher. Gathers repo context plus current official guidance so the coordinator can recommend the best next action before planning or implementation.
Agent Claude Code
Reviewer — Audits implementation for logic correctness, lint compliance, convention adherence, and test coverage gaps.
Agent Claude Code
Scout — Lightweight codebase reconnaissance agent. Read-only. Identifies structure, patterns, and hotspots without modifying files.
Agent Claude Code
Security Reviewer — OWASP-focused security audit. Identifies vulnerabilities, authentication flaws, injection risks, and data exposure. Read-only — does not modify code.
Agent Claude Code
This document defines the structure and format of the Technical Requirements Document (TECHREQ) generated by the prdanalyst agent as its primary output artifact.
Agent Claude Code
Test Planner — A specialized subagent invoked exclusively by utwriter and itwriter. Analyzes the codebase and generates or appends to the TestPlan.md document. Never invoked directly by the user.
Agent Claude Code
Unit Test Writer — Generates hand-written fakes, mock data objects, and unit tests for repositories, ViewModels, and use cases.
Agent Claude Code
Zero-trust validation agent. Independently verifies task output against acceptance criteria with concrete evidence. Reports per-criterion PASS/FAIL. Review depth tiers: full, standard, lightweight.
Agent Claude Code
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
Agent Claude Code
Compare two outputs WITHOUT knowing which skill produced them.
Agent Claude Code
Evaluate expectations against an execution transcript and outputs.
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: