sdd-verify

Verifies SDD change implementations against specs, design, and tasks with executed test evidence.

Updated Aug 5, 2026
One-click install
npx skills add https://github.com/Alountk/Bloodbowl_Project --skill sdd-verify-alountk
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sdd-verify
Source: https://github.com/Alountk/Bloodbowl_Project/tree/main/.opencode/skills/sdd-verify
Command: npx skills add https://github.com/Alountk/Bloodbowl_Project --skill sdd-verify-alountk

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It closes the gap between claimed and proven completion in spec-driven development by independently verifying that an implemented change actually matches its proposal, specs, design, and task list, backed by real test execution rather than static inspection alone. ## Core Features & Use Cases - Spec Compliance Matrix: Maps every requirement and scenario to a covering test with COMPLIANT, FAILING, UNTESTED, or PARTIAL statuses. - Runtime Evidence Enforcement: Runs test, build, and coverage commands and records exit codes and output hashes in a strict YAML envelope. - Strict TDD Auditing: Optionally validates RED/GREEN/TRIANGULATE evidence, test layer distribution, changed-file coverage, and assertion quality (tautologies, ghost loops, mock-heavy tests). - Use Case: After an apply phase finishes a change, the orchestrator delegates to this skill, which reruns the suite, cross-checks 144 data rows against spec scenarios, and returns a PASS, PASS WITH WARNINGS, or FAIL verdict with CRITICAL/WARNING/SUGGESTION issues. ## Quick Start Ask the orchestrator to run the SDD verification phase for the current change so it executes the tests and produces a verification report proving the implementation matches the specs.

Frequently Asked Questions about sdd-verify

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify an SDD change implementation against its specs?▼

Delegate to the sdd-verify sub-agent after all tasks are complete. It reads the proposal, specs, design, and tasks, runs the test and build commands, maps each spec scenario to a passing covering test, and returns a verification report with a PASS, PASS WITH WARNINGS, or FAIL verdict.

What does the SDD verification report contain?▼

The report contains a YAML envelope with verdict, exit codes, and output hashes, plus completeness, build and test evidence, a spec compliance matrix, correctness and design coherence tables, and issues grouped as CRITICAL, WARNING, or SUGGESTION.

Can verification pass if some tasks are still unchecked?▼

No. Any incomplete task blocks full verification and is flagged CRITICAL; the skill returns blocked without running the full suite. Unchecked tasks remain CRITICAL even when other artifacts are missing or warnings-only.

Does sdd-verify fix the issues it finds?▼

No. It only reports issues for the orchestrator or user to resolve. The skill acts as an independent quality gate and never modifies implementation code, fixes failing tests, or starts correction loops.

What happens in Strict TDD mode during verification?▼

The skill loads the strict-tdd-verify module and additionally audits the TDD Cycle Evidence table, classifies tests by layer, measures changed-file coverage, and runs an assertion quality audit that flags tautologies, ghost loops, and mock-heavy tests.

Why is static code inspection not enough for spec compliance?▼

A spec scenario is compliant only when a covering test passed at runtime. Source inspection alone cannot prove behavior, so the skill always executes the relevant test and build commands and records their exit codes and output hashes as evidence.