Explore

Use when 999.x backlog items need triage, or the user asks to review, promote, or prune the backlog. Shows each item with its accumulated context, promotes chosen ones into the active milestone, and removes stale ones.

Use when you resume work and the last session ended mid-phase, a .continue-here checkpoint exists, or a PLAN has no SUMMARY. Restores STATE.md context, shows status, and routes to the next command.

Use when you want phase research without planning: to investigate before deciding a phase is feasible, or to re-research after a plan already exists. Standalone command; most workflows should use /gsd-plan-phase instead.

Use when a finished GSD workspace and its worktrees should be deleted. Confirms first, runs git worktree remove for each member repo, and refuses if any repo has uncommitted changes.

Use when a planned but unstarted phase should come off the roadmap. Removes the phase, renumbers the phases after it into a clean sequence, and commits the change as the historical record.

Use after a GSD update, when local modifications were backed up to gsd-local-patches/ and must be merged back. Three-way compares the pristine baseline, the backup, and the new install, flagging conflicts instead of skipping files.

Use when the user wants a small, ad-hoc change handled with GSD guarantees (atomic commits, STATE.md) but without the full pipeline. Tasks land in .planning/quick/; --full, --discuss, --research, --validate add the skipped agents back.

Use when the user asks where the project stands, what to do next, or to resume after a break. Summarizes recent work and what is ahead, then routes to executing an existing plan or planning the next one.

Use when the user asks to profile how they work, adapt assistant responses to them, or refresh an existing developer profile. Generates USER-PROFILE.md from session analysis or a questionnaire.

Use when opening a pull request and the diff is cluttered with .planning/ commits. Creates a clean branch filtered of GSD planning artifacts so reviewers see only code changes.

Capture a forward-looking idea with trigger conditions — surfaces automatically at the right milestone

Use when a milestone audit reported gaps and you want the fix phases created in one pass. Reads MILESTONE-AUDIT.md, groups gaps into phases, adds them to ROADMAP.md and offers to plan each one.

Use when the user wants to pause work mid-phase or hand it off to a new session. Writes .continue-here.md with position, completed and remaining work, decisions and blockers, commits it as WIP, and prints resume instructions.

Use when the user says note this, jot this down, asks to list my notes, or wants a note promoted to a todo. Appends a timestamped note file in one Write with one confirmation line, no questions.

Use to keep moving without inspecting progress, or when the user asks what is next in the GSD workflow. Reads STATE.md, ROADMAP.md and phase dirs to invoke the next step; --force skips safety and verification gates.

Use when you need an isolated copy of one or more repos with its own .planning/, or want to work on several repos in parallel. Creates worktrees or clones plus WORKSPACE.md; then run /gsd-new-project inside.

Use when starting GSD in a repo that has no .planning/ yet, or the user wants a new project set up. Questioning, optional research, requirements, and roadmap produce .planning/PROJECT.md, REQUIREMENTS.md, ROADMAP.md, and STATE.md.

Use when the current milestone has shipped and it is time to scope the next one, or when the user asks to start a new milestone. Updates PROJECT.md, then runs questioning, research, requirements, and roadmap.

Use when onboarding someone to a finished milestone or preparing a team review. Reads ROADMAP, REQUIREMENTS, CONTEXT, SUMMARY, and VERIFICATION artifacts into one MILESTONE_SUMMARY in .planning/reports/, then offers Q&A.

Use when onboarding to an unfamiliar codebase, before a major refactor, on a brownfield project before /gsd-new-project, or to refresh .planning/codebase/. Spawns up to 4 parallel mapper agents that write 7 documents.

Use when the user asks which GSD workspaces exist, or where their isolated work lives. Scans the workspaces root for WORKSPACE.md manifests and prints name, path, repo count, strategy and project status.

Use when the user wants to see what the agent assumes about a phase's approach, order, scope, risks and dependencies before planning starts, instead of running a full discuss session. Read-only; ends with a prompt for feedback.

Use when the user asks where to get help, how to join the GSD community, or where to report a problem. Displays the Discord invite link and what the server is for.

Use when the user asks what the codebase intel knows, or runs /gsd-intel query, status, diff, or refresh. Queries or rebuilds the .planning/intel/ files, and stops with a disabled message when intel is off in .planning/config.json.

Use when urgent work surfaces mid-milestone and must land between two planned phases without renumbering the roadmap. Inserts it as a decimal phase such as 72.1 after the phase the user names.

Use when the user asks what GSD commands exist or how to use GSD. Outputs the complete GSD command reference verbatim, with no project analysis, status, or next-step suggestions.

Use when a GSD run failed, stalled, or produced the wrong result and the cause is unclear. Reads git history and .planning/ state and phase artifacts to diagnose root cause, then writes a report to .planning/forensics/.

Use for a trivial task the user can describe in one sentence: a typo, config change, small refactor or forgotten commit. Executes inline with no subagents and no PLAN.md; anything needing research or planning belongs to /gsd-quick.

Use when the user wants to execute a phase's plans, resume the remaining plans, or run only one wave. Discovers plans, groups them into dependency waves, spawns a fresh subagent per plan, then verifies the phase.

Use when the user asks to generate, update, refresh or verify project documentation, or to check that docs still match the codebase. Writes up to 9 doc files, each verified against real paths; supports --force and --verify-only.

Use when you know what you want but not which /gsd-* command to run. Routes freeform text to the best matching GSD command, confirms the match, then hands off without doing the work itself.

Use when a phase has unresolved decisions before planning and no CONTEXT.md exists. Asks adaptive questions on the phase's gray areas and writes {phase_num}-CONTEXT.md with locked decisions; --auto, --chain, --power change the questioning.

Use when a bug, failing test, or unexpected behaviour needs investigation, especially across sessions or context resets. Isolates it in a debugger subagent with persistent .planning/debug state, unlike skills/systematic-debugging.

Use when every phase in a milestone is complete and the user wants to archive it, tag the release, and start the next version. Archives roadmap and requirements to .planning/milestones/, updates PROJECT.md, then tags the version.

Use after gsd-code-review has written REVIEW.md, when you want its Critical and Warning findings fixed. Spawns the fixer agent, commits each fix atomically and writes REVIEW-FIX.md (--all, --auto to iterate).

Use when the user asks what is pending, what to do next, or to list todos (optionally filtered by area). Lists pending todos, loads full context for the selected one, and routes to work now, add to a phase, brainstorm, or create a phase.

Use when the user says run the rest autonomously, overnight, or hands-off through the remaining phases. Runs discuss, plan, execute per phase, pausing only for user decisions, then milestone audit, complete, and cleanup.

Use when the user asks what UAT or verification items are still outstanding across phases, or wants a prioritized manual test plan. Scans every phase for pending, skipped, blocked and human_needed items and flags stale docs.

Use before archiving a milestone, or when the user asks whether a milestone met its definition of done. Aggregates phase VERIFICATION.md files, tech debt and deferred gaps, then runs a cross-phase integration check.

Use after /gsd-audit-uat when the user wants the findings fixed, not just listed. Classifies each issue as auto-fixable or manual and fixes the auto-fixable ones with test verification and atomic commits; --severity sets the floor.

Use before /gsd-manager to fill in missing Depends on fields in ROADMAP.md. Analyzes file overlap, semantic dependencies and data flow between phases, then suggests dependency updates to confirm.

Use when an idea, task, or issue surfaces mid-session and the user wants it captured as a todo instead of doing it now. Infers the area, checks duplicates, and writes the todo file plus a STATE.md update.

Use when a completed phase needs unit or E2E tests, or the user asks to add tests for phase N. Classifies implementation files into TDD, E2E, or skip, gets plan approval, then generates RED-GREEN tests from SUMMARY.md and UAT criteria.

Use when the user asks to add a phase, feature, or scope to the end of the current milestone roadmap. Adds the next sequential phase directory, updates ROADMAP.md, and tracks the change in STATE.md.

Use when the user wants to park an idea or feature request that is not ready for active planning, or says add it to the backlog. Writes a 999.x entry to ROADMAP.md and creates its phase directory for later promotion.

Use when the user says 'cognee', asks about knowledge graph or graph memory, wants to switch memory backends, or needs a status check, dataset init, or graph view. Covers Cognee endpoints plus the Brain fallback.

Modify Cursor/VSCode user settings in settings.json. Use when the user wants to change editor settings, preferences, configuration, themes, font size, tab size, format on save, auto save, keybindings, or any settings.json values.

Convert 'Applied intelligently' Cursor rules (.cursor/rules/*.mdc) and slash commands (.cursor/commands/*.md) to Agent Skills format (.cursor/skills/). Use when the user wants to migrate rules or commands to skills, convert .mdc rules to SKILL.md format, or consolidate commands into the skills directory.

Create custom subagents for specialized AI tasks. Use when the user wants to create a new type of subagent, set up task-specific agents, configure code reviewers, debuggers, or domain-specific assistants with custom prompts.

Create Cursor rules for persistent AI guidance. Use when the user wants to create a rule, add coding standards, set up project conventions, configure file-specific patterns, create RULE.md files, or asks about .cursor/rules/ or AGENTS.md.

Multi-agent communication for AI coding tools. Agents message, watch, and spawn each other across terminals. Use when setting up hcom or troubleshooting.

智能配图与 PPT 信息图生成器。支持三种模式：(1) 文章配图模式 - 分析文章内容，生成插图；(2) PPT/Slides 模式 - 生成批量信息图；(3) Cover 模式 - 生成封面图。所有模式默认生成图片，`--prompt-only` 只输出 prompt。支持 Bento Grid 功能展示图风格（--style bento）。触发词：配图、插图、PPT、slides、封面图、thumbnail、cover、bento grid、功能展示图、feature showcase。

Vulnerability review, threat modeling, OWASP patterns, and secure coding assessment. Use when reviewing code security, designing secure systems, performing threat analysis, or validating security implementations.

Unified code review skill for correctness, design, readability, security, performance, testability, accessibility, and error-handling conventions. Use when reviewing changes, enforcing quality standards, or identifying technical debt.

Unified platform operations guidance for CI/CD pipeline design, deployment strategies, observability, SLI/SLOs, and incident-ready rollouts. Use when building release workflows, production monitoring, or reliability controls.

System architecture patterns including monolith, microservices, event-driven, and serverless, with C4 modeling, scalability strategies, and technology selection criteria. Use when designing system architectures, evaluating patterns, or planning scalability.

REST and GraphQL API design patterns, OpenAPI/Swagger specifications, versioning strategies, and authentication patterns. Use when designing APIs, reviewing API contracts, evaluating API technologies, or implementing API endpoints.

Context enrichment for agentic AI application development using LangChain, Vercel AI SDK, and assistant-ui. Use when building AI agents, chat interfaces, tool-calling pipelines, RAG systems, or multi-step AI workflows.

User research and insight synthesis for interviews, usability testing, personas, journey mapping, and actionable product/design recommendations. Use when planning studies, collecting evidence, synthesizing patterns, or converting findings into prioritized decisions.

Requirement gathering techniques, stakeholder analysis, user story patterns, and specification validation. Use when clarifying vague requirements, resolving conflicting needs, documenting specifications, or validating requirements with stakeholders.