Creates .specs/ directory, writes CONSTITUTION.md skeleton, and initializes the state machine. Call this first before any other SDD tool.

Returns 7 structured discovery questions tailored to your project idea. Covers: scope, users, constraints, integrations, performance, security, and deployment.

Generates and writes SPECIFICATION.md with all requirements in EARS notation. Validates each requirement against EARS patterns.

Reads SPECIFICATION.md and returns up to 5 disambiguation questions targeting ambiguous or incomplete requirements.

Generates and writes DESIGN.md with architecture overview, Mermaid diagrams, ADRs, and API contracts.

Generates and writes TASKS.md with pre-implementation gates, sequenced tasks with [P] parallel markers, effort estimates, and requirement traceability.

Reads all spec files, generates ANALYSIS.md with traceability matrix and coverage report, and returns a gate decision (APPROVE, CHANGES_NEEDED, or BLOCK).

Validates that the current phase's required files exist, then transitions the state machine to the next phase.

Compares specification requirements against implementation files and returns a drift report showing which requirements are implemented and which are missing.

Returns the current pipeline status including: current phase, completed phases, files on disk, completion percentage, and recommended next action.

Returns the raw Markdown template for a given artifact type with {{placeholder}} variables intact. Does not write any files.

Generates and writes BUGFIX_SPEC.md with current behavior, expected behavior, unchanged behavior, root cause analysis, and test plan. Not gated by the state machine.

Scans the workspace project structure and returns auto-steering context: detected language, framework, package manager, folder structure, and key files.

Appends an amendment entry to CONSTITUTION.md's changelog and updates the amendment_count in frontmatter.

Reports which external MCP servers are recommended for the full Specky experience. Shows what each server does, which Specky tools it enhances, and how to install it. Run this first to understand what integrations are available.

Parses a meeting transcript (VTT, SRT, TXT, or MD) and extracts structured data: participants, topics, decisions, action items, raw requirements, constraints, and open questions. Supports Teams, Zoom, Google Meet, and Otter.ai transcripts.

FULLY AUTOMATED: Reads a meeting transcript, extracts requirements, and runs the complete SDD pipeline in one call. Creates CONSTITUTION.md, SPECIFICATION.md, DESIGN.md, TASKS.md, and ANALYSIS.md from a single transcript file. Supports VTT (Teams), SRT (Zoom), TXT, and MD formats.

Scans a folder for transcript files (.vtt, .srt, .txt, .md) and runs the full SDD auto-pipeline for each one. Designed for Power Automate + OneDrive workflows where meeting transcripts are saved automatically to a shared folder. Each transcript becomes its own feature spec package.

Imports a document (PDF, DOCX, PPTX, TXT, MD, VTT, SRT) or raw text and converts it to Markdown for SDD processing. Returns the converted content, metadata, and word count.

Prepares a structured payload for extracting design context from a Figma file. The AI client should use the returned routing_instructions to call Figma MCP's get_design_context tool with the provided file key and node ID.

Scans a directory for supported documents (PDF, DOCX, PPTX, TXT, MD) and converts each to Markdown. Returns an array of conversion results with total count and per-file metadata.

Generates a domain-specific quality checklist (security, accessibility, performance, etc.) by analyzing SPECIFICATION.md and DESIGN.md. Writes CHECKLIST.md.

Reads TASKS.md and checks code_paths for implementation evidence. Detects phantom completions — tasks marked [x] but with no corresponding code. Writes VERIFICATION.md.

Validates specification and design against a compliance framework (HIPAA, SOC2, GDPR, PCI-DSS, ISO27001, or general). Writes COMPLIANCE.md.

Cross-artifact consistency analysis: checks alignment between SPECIFICATION.md, DESIGN.md, and TASKS.md. Finds orphaned requirements, missing designs, and untraced tasks. Writes CROSS_ANALYSIS.md.

Validates requirement statements against EARS notation patterns (ubiquitous, event-driven, state-driven, optional, unwanted, complex). Accepts a direct list of requirements OR reads from SPECIFICATION.md. Returns per-requirement compliance results with actionable suggestions.

Generates a single Mermaid diagram from a specification artifact. Supports 17 diagram types: flowchart, sequence, class, ER, state machine, C4 context, C4 container, C4 component, C4 code, activity, use case, DFD (data flow), deployment, network topology, Gantt, pie chart, and mind map.

Generates ALL diagram types for a feature in one call. Produces architecture, sequence, ERD, flow, dependency, and traceability diagrams from all available artifacts.

Generates user stories with acceptance criteria and flow diagrams from SPECIFICATION.md. Each story includes a Mermaid flowchart of the user journey.

Generates a FigJam-ready diagram payload from DESIGN.md. Returns structured data with routing_instructions for the AI client to call Figma MCP's generate_diagram tool.

Reads DESIGN.md to detect infrastructure needs and generates Terraform or Bicep files. Returns generated file contents, variables, and a Mermaid diagram of the infrastructure topology.

Generates a validation payload for Terraform MCP (plan/validate) or Azure MCP (template validation). The AI client routes this payload to the appropriate MCP server for execution.

Reads DESIGN.md or uses sdd_scan_codebase results to detect the tech stack, then generates a Dockerfile and optionally a docker-compose.yml. Supports multi-stage builds for smaller production images.

Detects the project tech stack and generates a Docker-based local development environment (Dockerfile + docker-compose.yml). Returns a payload with routing_instructions for Docker MCP to create and manage containers.

Detects the project tech stack and generates a devcontainer configuration suitable for GitHub Codespaces. Returns a payload with routing_instructions for GitHub MCP's create_codespace tool.

Generates .devcontainer/devcontainer.json from the detected tech stack and DESIGN.md. Writes the file to disk for local use with VS Code Dev Containers or GitHub Codespaces.

Generates a branch name following SDD conventions and returns a command_hint for creating the branch. Does not execute git commands — the AI client or user runs the command.

Transforms TASKS.md into platform-specific work item payloads (GitHub Issues, Azure Boards, or Jira). Returns routing_instructions for the AI client to create items via the appropriate MCP server.

Generates a pull request payload from SPECIFICATION.md and TASKS.md with spec summary, requirements covered, and task progress. Returns routing_instructions for GitHub MCP's create_pull_request tool.

Reads TASKS.md and produces an ordered implementation roadmap with phases, parallel groups, dependency resolution, and checkpoints. Does NOT write code — it generates the plan the developer or AI agent follows.

Takes an array of research questions, generates RESEARCH.md with structured entries (question, findings placeholder, sources, recommendation, status), and writes it to the feature directory.

Generates comprehensive feature documentation from SPECIFICATION.md, DESIGN.md, TASKS.md, and ANALYSIS.md. Writes a combined Markdown file to docs/ with all sections.

Extracts API endpoints from DESIGN.md and generates structured API documentation with request/response examples. Writes to docs/api-{feature}.md.

Generates an operational runbook with deployment, monitoring, troubleshooting, and rollback procedures. Writes to docs/runbook-{feature}.md.

Generates ALL documentation types in parallel: full docs, API docs, runbook, onboarding guide, and SDD journey. All documents are written to docs/ directory. This is the fastest way to generate complete project documentation.

Generates a developer onboarding guide with feature overview, architecture summary, getting started steps, key concepts, and file locations. Writes to docs/onboarding-{feature}.md.

Generate test stubs from acceptance criteria in SPECIFICATION.md and TASKS.md. Supports 6 frameworks: vitest, jest, playwright, pytest, junit, xunit. Each test stub traces to a requirement ID for full traceability.

Reads test results JSON and cross-references with requirement IDs from SPECIFICATION.md. Reports requirement coverage percentage, uncovered requirements, and a traceability matrix.

Creates a named snapshot of all spec artifacts (CONSTITUTION.md, SPECIFICATION.md, DESIGN.md, TASKS.md, etc.) and the current pipeline state. Use before making major changes so you can rollback if needed.

Restores all spec artifacts to a previous checkpoint snapshot. Overwrites current files with the checkpoint versions. Creates an automatic backup checkpoint of current state before restoring.

Lists all available checkpoints for a feature with their labels, dates, and phases.

Generates a complete EARS specification from a natural language feature description. Automatically extracts requirements, classifies into EARS patterns (ubiquitous, event-driven, state-driven, optional, unwanted), generates acceptance criteria, and identifies areas needing clarification. This is the fastest way to go from idea to spec.

Extracts universal properties (invariants, round-trips, idempotence) from EARS requirements and generates property-based tests using fast-check (TypeScript) or hypothesis (Python). Unlike example-based tests, PBT uses random input generation to discover edge cases that manual tests miss.

Generate a self-contained HTML metrics dashboard for a feature. Reads SPECIFICATION.md, ANALYSIS.md, VERIFICATION.md, CHECKLIST.md and .sdd-state.json. Produces metrics-dashboard.html with: requirement count, task coverage, compliance score, checklist pass rate, and a phase timeline with durations.

Return the full model routing decision table for all 10 SDD pipeline phases. Includes the optimal model, chat mode, extended thinking setting, rationale, evidence arXiv ID, and cost savings vs Opus-for-everything for your team size.

Return the context tier assignment (Hot/Domain/Cold) for all spec artifacts in the active feature. Includes estimated token load for current session vs universal loading, and savings percentage.

Check RBAC access for the current role. Returns the active role, whether a specific tool is accessible, and a summary of what each role can do. Useful for diagnosing permission issues in enterprise deployments.