Document-first agentic dev harness for Claude Code and Codex CLI — scaffolds charter, architecture, and a sub-agent workflow for a new project in one shot.
This repository also configures its own agents. See what agentic-dev-harness tells them →
Agent Claude Code
A data-measurement and interpretation adviser for product work. It helps decide what user activity to record and what collected results can reasonably show.
Agent Claude Code
Use proactively for major product or architecture decisions, milestone decomposition, and important design tradeoff analysis. Avoid routine low-impact tasks.
Agent Claude Code
Use proactively for scoped implementation work. Best for task-level coding, tests, and localized refactors that should stay within a documented workitem.
Agent Claude Code
A legal-advice agent for checking jurisdiction-specific rules, privacy policies, terms of service, and open-source licence compatibility using primary sources.
Agent Claude Code
Use for visual/UX design authoring — reference decomposition, design principles, divergent concept drafts, milestone screen prototypes, and DESIGN.md authoring support. Generation only; auditing stays with reviewer[design].
Agent Claude Code
A product-marketing agent for writing and planning the words users see inside a product, such as positioning, landing pages, pricing pages, onboarding, SEO structure, and emails.
Agent Claude Code
Use proactively for everyday planning, requirement cleanup, and decomposing work into milestones, features, and tasks. For major product or architecture decisions, prefer architect.
Agent Claude Code
Use proactively for QA sweeps, edge-case analysis, user-visible risk review, and regression checks after meaningful implementation changes.
Agent Claude Code
Use for gathering and distilling external information — official docs, primary sources, papers — when implementation or planning needs current, citable facts. Report-only; never edits code or docs.
Agent Claude Code
Use proactively for critical review of documents or code when you need contradictions, missing requirements, hidden complexity, or vague assumptions identified.
Agent Claude Code
A report-only adviser for security decisions made at the system-design level. It examines threats, data protection, authentication and authorization boundaries, and the technical requirements of regulations.
Agent Claude Code
A report-only adviser for business strategy. It examines revenue models, pricing tiers, unit economics, market sizing, competitive positioning, and the relationship between customer-acquisition cost and customer value.
Agent Claude Code
Use proactively after implementation to verify scope alignment, document consistency, obvious regression risk, and completion readiness.
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: