Skip to content

Instantly share code, notes, and snippets.

@kingargyle

kingargyle/sdd_test_prompt.md

Created March 12, 2026 01:38
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save kingargyle/f6750495322891a3e9a10b2f9d781d7e to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save kingargyle/f6750495322891a3e9a10b2f9d781d7e to your computer and use it in GitHub Desktop.
Download ZIP
Test Suite for SDD
Raw
sdd_test_prompt.md

Write me a spec that can be used to test out the functionality of the  @sdd_implementation_v1.md  or any general SDD implementation. The spec should generate a test suite that can be run to validate the functions of the spec implementation is working as it is designed or intended. It should be able to update whatever existing SDD test suite exists if the underly spec implementation changes.

License

MIT License Copyright (c) 2026 David Carver and NineWorlds

Raw
sdd_validation_suite.md

[SPEC] SDD Protocol Validation Suite

Goal

  • Create a standardized validation suite to ensure that any agent implementing the Phased SDD (Spec-Driven Development) protocol adheres to the requirements defined in prompts/plans/sdd_implementation_v1.md.
  • The suite should be able to update itself if the underlying SDD implementation version changes.

Context

  • Current State: The SDD protocol is defined in prompts/plans/sdd_implementation_v1.md (v2.1.0). Currently, validation is manual/vibe-based.
  • Constraints: Must use project-standard documentation and directory structures.
  • Files:
    • prompts/plans/sdd_implementation_v1.md
    • prompts/agents/SPEC_WRITER.md
    • prompts/templates/TASK_SPEC.md
    • AGENTS.md

Standards Compliance

  • Architecture: Phased GSD (Goal, Steps, Deliverables).
  • Organization: Plan-specific subdirectories with memory/ folders.
  • Testing: Behavior validation for AI agents.

Impact Surface

  • Dependencies: Any future updates to the SDD protocol will require a re-run of this validation suite.
  • Consumers: All AI agents (Cursor, Claude Dev, etc.) interacting with this repository.

Risk & Rollback

  • Potential Risks: Validation suite might become stale if the protocol is updated without updating the suite.
  • Rollback Strategy: Git revert of the prompts/plans/sdd_validation_suite/ directory.

Phase 1: Test Suite Definition

Tasks

  • Task 1.1: Define Scenario 1 - Complex Task Detection
    • Requirement: Agent must refuse to code and initiate a spec if > 2 files are affected.
    • Input: "Refactor the entire networking layer to use Ktor."
    • Expected Output: Refusal + initiation of a Phased GSD Spec.
  • Task 1.2: Define Scenario 2 - Directory & Memory Structure
    • Requirement: Agent must create prompts/plans/<name>/ and prompts/plans/<name>/memory/.
    • Input: Requesting a spec for a specific task.
    • Expected Output: Files created in the correct relative paths.
  • Task 1.3: Define Scenario 3 - Atomic Execution & Gates
    • Requirement: Agent must STOP and wait for human approval after each phase.
    • Expected Output: Presence of [WAIT FOR APPROVAL] or equivalent STOP markers in generated plans.

Verification

  • Manual verification of the test case definitions.

[WAIT FOR APPROVAL]

Phase 2: Validation Scripting/Checklist

Tasks

  • Task 2.1: Create validate_sdd.sh (Optional/Conceptual)
    • Action: Create a helper script or a markdown checklist that can be "run" by an agent to self-verify.
    • Content: A checklist that parses the current version of sdd_implementation_v1.md and checks for compliance in the last 3 generated plans.
  • Task 2.2: Implement Version Tracking
    • Action: Add a "Suite Version" that matches the "SDD Protocol Version".

Verification

  • Run the checklist against the sdd_validation_suite itself (dogfooding).

[WAIT FOR APPROVAL]

Phase 3: Self-Update Mechanism

Tasks

  • Task 3.1: Define Update Protocol
    • Action: Add instructions to SPEC_WRITER.md that if sdd_implementation_v1.md is updated, the sdd_validation_suite.md must be updated in the same PR/commit.

Verification

  • Simulate a protocol version bump and check if the agent proposes an update to the suite.

[WAIT FOR APPROVAL]

Final Deliverables

  • prompts/plans/sdd_validation_suite/sdd_validation_suite.md
  • prompts/plans/sdd_validation_suite/memory/validation_log.md

Discovery & Learning Log

  • Initial creation of the validation suite to prevent protocol drift.

Cleanup & Manual Actions (Human Required)

  • Review and approve the validation scenarios.

License

MIT License Copyright (c) 2026 David Carver and NineWorlds

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment