SDD Spec Architect

    1

    Architects software specifications, architecture decisions, and dependency-ordered tasks from ambiguous requests.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    SDD Spec Architect

    Example session with this skill installed

    Create a specification for adding a JWT-based authentication middleware to our existing Express.js API. Check the /src folder for current patterns.

    • Read your context and instructions
    • Compiled the sdd spec architect

    I have analyzed the /src directory. The project currently uses a custom ErrorHandler in middleware/ but lacks a session or token validation layer. I have drafted the technical specification below.

    Technical Specification: JWT Middleware

    IDRequirementAcceptance Criteria
    REQ-01Token ExtractionExtract Bearer token from Authorization header; return 401 if missing.
    REQ-02Signature ValidationVerify token against process.env.JWT_SECRET using jsonwebtoken library.
    REQ-03Context InjectionAttach decoded payload to req.user for downstream route access.
    REQ-04Expiry HandlingReject tokens that have passed their exp claim with a specific 'Token Expired' message.

    Architecture Decisions

    • Library: Using jsonwebtoken for industry-standard compatibility.
    • Strategy: Stateless JWT validation to maintain horizontal scalability of the API nodes.
    • Failure Path: Unauthorized requests will be passed to the existing ErrorHandler via next(err).

    Implementation Tasks

    1. [TASK-01] Install jsonwebtoken and @types/jsonwebtoken. (Dep: None)
    2. [TASK-02] Create src/middleware/auth.ts implementing REQ-01 through REQ-04. (Dep: TASK-01)

    [TASK-03] Create unit tests in tests/middleware/auth.test.ts for valid, expired, and malformed tokens. (Dep: TASK-02)
    4. [TASK-04] Apply middleware to protected routes in src/routes/user.ts. (Dep: TASK-03)

    Verification (NOT RUN)

    • npm test tests/middleware/auth.test.ts
    • Manual curl: curl -H "Authorization: Bearer <invalid_token>" http://localhost:3000/api/profile

    Next steps

    1. Confirm the list of protected route files.
    2. Provide the JWT_SECRET for the local .env test environment.
    3. Approve this spec to proceed with task implementation.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Software projects often stall due to ambiguous requirements and unplanned technical debt. This skill acts as a Specification-Driven Development (SDD) architect to bridge the gap between high-level ideas and executable code. It enforces a rigorous pre-coding phase where system design, edge cases, and dependency-ordered tasks are defined before a single line of logic is written.

    What it does

    • Requirement Analysis: Converts vague objectives into numbered functional and nonfunctional requirements.
    • System Architecture: Proposes data contracts, failure paths, and rollback strategies while recording architectural decisions.
    • Task Orchestration: Generates dependency-ordered implementation tasks mapped directly to requirement IDs.
    • Traceability Auditing: Ensures every requirement has at least one measurable acceptance criterion and verification step.
    • Risk Assessment: Identifies blockers, security considerations, and unknowns that require stakeholder approval.

    How it works

    1. Context Analysis: The agent scans the repository, project conventions, and existing modules to distinguish facts from assumptions.
    2. Scoping: It defines goals, non-goals, and constraints while surfacing targeted questions for any discovered ambiguities.
    3. Spec Drafting: The agent builds the technical specification, including architecture decisions and measurable acceptance criteria.
    4. Tasking: It breaks down the plan into prioritized tasks with inputs, outputs, and verifiable completion triggers.

    Frameworks & tools

    This skill is framework-agnostic and works with any stack. It is designed for use within agentic IDEs like Claude Code, Cursor, or Windsurf that support SKILL.md definitions and local repository context.

    Why this beats prompting it yourself

    Standard prompts often skip the "audit" phase, leading to missing edge cases or circular dependencies. This skill enforces a structured SDD workflow that separates planning from execution, ensuring every task is grounded in verified repository context rather than AI hallucinations.

    Use cases

    • Translate a messy PRD or Slack thread into a formal technical design document.
    • Map out complex feature migrations where dependency ordering is critical.
    • Create testable acceptance criteria for QA and engineering alignment.
    • Audit existing project context to identify technical risks before starting a sprint.

    Known limitations

    The skill requires clear user objectives and repository access to be effective. It will not execute code or perform destructive operations like git merges without explicit separate authorization.

    Requires clear initial objectives. Does not execute code or perform deployments. Estimates are only provided if local evidence supports them. Environment-specific CLI tools must be pre-installed.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Permissions

    Read Files
    Network Access
    Listed1 month ago
    Updated13 days ago

    What's inside

    Frequently Asked Questions

    The skill converts ambiguous requests into structured documentation, including numbered requirements, data contracts, and dependency-ordered tasks, preventing technical debt by ensuring architectural decisions are finalized before coding begins.

    This skill is framework-agnostic and designed for agentic IDEs such as Claude Code, Cursor, and Windsurf that support `SKILL.md` definitions and local repository context.

    Unlike a simple prompt, this skill performs context-aware audits and risk assessments by scanning your existing repository to ensure every task is grounded in project reality rather than AI hallucinations.

    You will receive the `SKILL.md` definition file and a configuration guide that enables your AI agent to follow the Specification-Driven Development (SDD) framework within your local environment.

    No, the architect focuses on planning and task orchestration; it defines the roadmap and verification steps, but execution of the code or git operations is handled separately by your coding agent.

    New in Agents & Orchestration