AI Engineering

Execute Phase runs all plans in a phase using wave-based parallel execution with subagents. It discovers plans, analyzes dependencies, groups them into waves, and collects results while preserving checkpoints and verification gates.

GSD Config configures GSD settings — workflow toggles, advanced knobs, integrations, and model profile — for Claude Code. Use it to switch model profiles, manage API keys, and tune planning and execution settings.

GSD Discuss Phase gathers phase context through adaptive questioning before planning. It loads prior context, scouts the codebase, identifies undecided gray areas, and produces a CONTEXT.md with implementation decisions for downstream agents.

GSD Explore runs a Socratic ideation session that probes an idea with questions and optionally spawns research, then routes the outputs to GSD artifacts like notes, todos, seeds, research questions, requirements, or phases.

GSD Extract Learnings pulls decisions, lessons, patterns, and surprises from completed phase artifacts into a LEARNINGS.md file. Use it after a GSD phase completes to capture retrospective insights from PLAN.md, SUMMARY.md, VERIFICATION.md, UAT.md, and STATE.md.

GSD Graphify builds, queries, and inspects the project knowledge graph stored in .planning/graphs/. Use it to build or rebuild the graph, search for terms, check status and diff, or see changes since last build.

GSD Health validates .planning/ directory integrity and reports actionable issues. Use --repair to fix problems and --context to check the current session's context utilization.

Ad Image Generator creates paid-ad image assets from a validated creative brief and brand profile using an explicitly configured image provider. Use it to generate campaign visuals after the brief and brand profile are validated, with audit-ready manifests and policy checks.

Plannotator Setup Goal turns an idea or objective into a goal package by interviewing the user, building a reviewed fact sheet, and exploring the codebase to produce an execution plan.

Brainstorming turns ideas into design specifications through collaborative questioning before any implementation work. Use it before creating features, building components, adding functionality, or modifying behavior to explore user intent, requirements, and design.

Dispatching Parallel Agents delegates independent debugging tasks to parallel subagents with isolated contexts, so each can work on a separate problem without interference. Use it when multiple unrelated test failures or subsystem issues can be investigated concurrently.

Executing Plans loads a written implementation plan, critically reviews it, executes all tasks with verification and commits, and reports completion. Use when you have a plan with review checkpoints and need to execute it in a separate session.