Execute the implementation planning workflow using the plan template to generate design artifacts.
Generates implementation plans by following a structured workflow. Use this when you need to design features, resolve technical unknowns, and create artifacts like data models and API contracts before coding.
/plugin marketplace add nprbst/speck/plugin install speck@speck-market$ARGUMENTS
You MUST consider the user input before proceeding (if not empty).
Parse command-line flags from user input:
--stacked: Enable stacked PR workflow mode (write workflow metadata to plan.md)Setup: Extract prerequisite context from the auto-injected comment in the prompt:
<!-- SPECK_PREREQ_CONTEXT
{"MODE":"multi-repo","FEATURE_DIR":"/path/to/root-repo/specs/010-feature","IMPL_PLAN":"/path/to/child-repo/specs/010-feature/plan.md","TASKS":"/path/to/child-repo/specs/010-feature/tasks.md","REPO_ROOT":"/path/to/child-repo","TEMPLATE_DIR":"/path/to/plugin/templates","AVAILABLE_DOCS":["../../../8-specs/specs/010-feature/spec.md",".speck/memory/constitution.md"]}
-->
Path Usage:
FEATURE_DIR: Directory containing shared artifacts (spec.md, research.md, data-model.md) - READ ONLYIMPL_PLAN: Full path where plan.md should be written - WRITE HERETASKS: Full path where tasks.md should be written (used by /speck.tasks)REPO_ROOT: Root directory of current repository (for relative path calculations)TEMPLATE_DIR: Directory containing templates (plan-template.md, spec-template.md, etc.) - USE FOR TEMPLATESMODE: "single-repo" or "multi-repo" (child in multi-repo setup)Multi-repo behavior:
Fallback: If the comment is not present (VSCode hook bug), run:
speck setup-plan --json
Parse JSON for FEATURE_SPEC, IMPL_PLAN, SPECS_DIR, BRANCH.
IMPORTANT: Do NOT invent or guess CLI paths. Use ONLY the path above.
Parse JSON output to extract FEATURE_SPEC, IMPL_PLAN, SPECS_DIR, and BRANCH.
Load context (use Read tool for all files):
{TEMPLATE_DIR}/plan-template.md (e.g., /Users/.../.claude/plugins/.../templates/plan-template.md)Execute plan workflow: Follow the structure in IMPL_PLAN template to:
**Workflow Mode**: stacked-pr## User Story Groupings
**Suggested Stacking Strategy**:
- Branch 1: US1, US2 (Database layer)
- Branch 2: US3, US4 (API endpoints)
- Branch 3: US5, US6 (UI components)
These groupings represent natural boundaries for stacked PRs. Each group can be implemented in a separate branch with its own pull request.
Write plan.md: Use the IMPL_PLAN path from prerequisite context to write plan.md file.
Stop and report: Command ends after Phase 2 planning. Report branch, IMPL_PLAN path, and generated artifacts.
Extract unknowns from Technical Context above:
Generate and dispatch research agents:
For each unknown in Technical Context:
Task: "Research {unknown} for {feature context}"
For each technology choice:
Task: "Find best practices for {tech} in {domain}"
Consolidate findings in research.md using format:
{REPO_ROOT}/specs/{feature}/research.md (local to child repo - each repo does its own research)Output: research.md with all NEEDS CLARIFICATION resolved
Prerequisites: research.md complete
Before generating contracts: Check for existing shared contracts
If {FEATURE_DIR}/contracts/ exists and contains files:
If no contracts exist, proceed to generate them below
Extract entities from feature spec → data-model.md:
{REPO_ROOT}/specs/{feature}/data-model.md (local to child repo - each repo has its own data model)Generate API contracts from functional requirements (only if no shared contracts exist):
{FEATURE_DIR}/contracts/ (shared - goes to root repo, accessible to all child repos)Generate quickstart.md with developer setup instructions:
{REPO_ROOT}/specs/{feature}/quickstart.md (local to child repo)Agent context update:
speck update-agent-contextOutput: data-model.md (local), contracts/* (shared, if created), quickstart.md (local), agent-specific file