skills.rest
nebius
nebiusOFFICIAL
Explore all skills by this author →
WebsiteGitHub

sdlc-workflow-test

Verifies the Agentic SDLC workflow through deterministic checks and an optional three-tier Docker live run.

7|5|Updated Jul 22, 2025
Search Tags:#docker#sdlc#test harness#workflow verification#evidence manifest#gui uat#hook parity
One-click install
npx skills add https://github.com/nebius/nebius-ps-services --skill sdlc-workflow-test-nebius
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sdlc-workflow-test
Source: https://github.com/nebius/nebius-ps-services/tree/main/skills/sdlc-workflow-test
Command: npx skills add https://github.com/nebius/nebius-ps-services --skill sdlc-workflow-test-nebius

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Teams adopting the Agentic SDLC workflow need a safe way to confirm that all phase skills, hooks, and execution capabilities actually work as designed before trusting them on a real repository. This Skill acts as an external test harness that audits the SDLC system without joining the workflow itself. ## Core Features & Use Cases - Deterministic preflight verification: Runs static discovery, source-installed skill parity, hook payload comparison, and capability regression checks against a disposable nested Git fixture. - Optional three-tier live profile: With explicit --create, builds one real local task-board GUI, Django/Gunicorn server, and PostgreSQL database through the normal SDLC workflow, performs computer-use GUI UAT, then destroys every exactly owned Docker resource. - Lifecycle management: Supports --create --keep, --resume, and standalone --destroy with fail-closed ownership labels, immutable verification-ID fencing, and idempotent ALREADY_DESTROYED results. - Use Case: After changing the SDLC design doc or hook source, run the no-flag verifier to get a PASS/PARTIAL/FAIL report confirming whether the workflow is safe to try on a low-risk repository. ## Quick Start Ask the agent to run the sdlc-workflow-test skill with no flags to verify the Agentic SDLC workflow and produce a readiness report.

Frequently Asked Questions about sdlc-workflow-test

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

FAQPage Schema
How do I verify the Agentic SDLC workflow is working?▼

Invoke the skill with no lifecycle flags to run the lightweight deterministic verifier, which checks skill parity, hooks, and capability regressions against a disposable fixture. It writes a PASS, PARTIAL, or FAIL report to the verification root without touching real repositories.

What does the --create flag do in the SDLC workflow test?▼

The --create flag runs preflight, destroys any previous owned test environment, then builds one real three-tier application (task-board GUI, Django/Gunicorn, PostgreSQL) through the normal SDLC workflow. It performs computer-use GUI UAT and cleans up all exactly owned Docker resources even after failure.

Does the SDLC workflow test work with both Codex and Claude?▼

Yes, the skill supports both hosts through their native invocation syntax and isolated native home directories. Hook parity checks cover Codex hooks.json/config.toml and Claude settings.json registrations, and resume or cleanup requires the originally recorded host.

Can the verifier modify my installed skills or hooks?▼

No, the verifier is explicitly forbidden from editing installed skills, hook configuration, credentials, or real project trees. All writes are confined to a private verification root and a disposable nested Git project with no remote.

Why does the verification report show PARTIAL instead of PASS?▼

PARTIAL appears when live evidence is missing, optional hook registration is absent, or required tooling such as Computer Use is unavailable. PASS is fail-closed and requires every deterministic and live lane to pass with digest-backed evidence.

How do I clean up a retained three-tier test environment?▼

Run the skill with --destroy to remove the one retained owned application, containers, volumes, images, and raw evidence while preserving sanitized reports. Repeating the command returns ALREADY_DESTROYED, and ambiguous ownership fails closed without deleting anything.

Related Skills

Explore More Software Engineering Skills→
DagimAlemayehuuDagimAlemayehuu

project-setup-alignment

Audit and align repositories into the agentic SDLC workflow without destructive changes.

Community
Advanced
etrigan16etrigan16

sdd-onboard

Guides users through a complete spec-driven development cycle on a real codebase.

Community
Advanced
josixjosix
7

skill-agent-mapping

Resolve agent ownership and consumption of skills in multi-agent orchestration.

Community
Basic
AINative-StudioAINative-Studio

database-query-best-practices

Prevent PostgreSQL connection pool exhaustion in local Railway queries.

Official
Intermediate