lhpaul/ai-dev-framework-template
Agent Claude Code
Run the automated reviewer loop (and CI loop) for a PR. Use when the user wants to run the reviewer loop on a specific PR or the current branch's PR until it is ready for human review or escalated.
A tool-agnostic AI-assisted development framework template. Structured workflow (Spec → Plan → Implement → Review) with canonical protocols, agent definitions for Claude Code/Cursor, and a guided project setup.
lhpaul/ai-dev-framework-template
Agent Claude Code
Run the automated reviewer loop (and CI loop) for a PR. Use when the user wants to run the reviewer loop on a specific PR or the current branch's PR until it is ready for human review or escalated.
lhpaul/ai-dev-framework-template
Agent Claude Code
Development review stage. Use when an implementation PR needs review against the spec, plan, and best practices. Applies fixes directly for blocking and important issues. Reports issues requiring human/product decisions.
lhpaul/ai-dev-framework-template
Agent Claude Code
Design review stage. Use when an implementation PR includes frontend file changes. Launches a browser via the configured browserautomation.provider, renders affected pages or components, captures screenshots, checks for console errors, and runs an axe-core accessibility check. Posts a structured PR comment with the…
lhpaul/ai-dev-framework-template
Agent Claude Code
In Development stage. Handles four paths — Full Pipeline (feature with spec+plan), Refactor (code restructuring with plan only, no spec), Fast Track (bug or simple change, no spec/plan needed), and Hotfix (critical production bug from main). Implements code, verifies build/lint/tests, opens PRs, and resolves reviewer…
lhpaul/ai-dev-framework-template
Agent Claude Code
Plan review stage. Use when an implementation plan branch or PR needs review for spec alignment, completeness, and feasibility. Reads the spec, plan, and codebase to validate the approach, and can push reviewer-loop fixes.
lhpaul/ai-dev-framework-template
Agent Claude Code
Coordination agent for a single workflow item via /run-item (or deprecated /run-item-work). Runs the shared bounded prelude then Protocol 91 until waiting on a human, blocked, or escalated.
lhpaul/ai-dev-framework-template
Agent Claude Code
Batch orchestration agent. Discovers what can advance or start, proposes the largest safe batch by priority and parallelization feasibility, dispatches approved item work, and supervises the batch until each item is waiting on a human, blocked, or escalated.
lhpaul/ai-dev-framework-template
Agent Claude Code
Post-merge QA on develop or develop- . Confirmed scope, environment preflight, flow/UX exercise, optional design-asset fidelity, one fix PR for safe defects (no backlog item).
lhpaul/ai-dev-framework-template
Agent Claude Code
Spec Ready stage. Use when a new feature needs a spec written. Conducts a structured alignment conversation with the human, then writes the feature spec, runs its reviewer gate, and resolves PR readiness. Do NOT use for bugs or simple changes (use the developer agent with fast track instead) or for refactors (use the…
lhpaul/ai-dev-framework-template
Agent Claude Code
Onboarding agent. Run this when first setting up a new project from this template. Guides a structured conversation to generate all project-specific documentation (business domain, architecture, tech stack, best practices) and configures the framework for your project.
lhpaul/ai-dev-framework-template
Agent Claude Code
Retrospective analysis agent. Analyzes completed work (a batch or individual item) to identify process improvement opportunities, presents them to the human, and executes the chosen action for each. Use after a batch or item run to surface workflow improvements, or invoke on-demand with a PR number, branch, or date as…
lhpaul/ai-dev-framework-template
Agent Claude Code
Smoke test stage. Use when a code review has been approved and the implementation needs manual smoke testing before merge. Executes the smoke test runbook using browser automation.
lhpaul/ai-dev-framework-template
Agent Claude Code
Spec review stage. Use when a spec branch or PR needs review for completeness, clarity, and testability. Applies fixes directly where possible, can push reviewer-loop fixes, and reports issues requiring human input.
lhpaul/ai-dev-framework-template
Agent Claude Code
Plan Ready stage. Use when an implementation plan needs to be written. For Full Pipeline items, a spec must be approved first. For Refactor items, the work item brief replaces the spec. Reads the codebase, resolves technical approach questions, then writes the implementation plan, runs its reviewer gate, and resolves…