Overview

/cicd-unified is a production command in the DevOps category of the Prismatic Platform that provides a single entry point for all continuous integration and continuous deployment workflow actions. Instead of managing separate commands for pipeline triggering, status monitoring, artifact retrieval, and deployment control, this command unifies the entire CI/CD lifecycle under one interface. It abstracts away the differences between GitLab CI pipeline syntax, deployment targets (Fly.io staging and production), and quality gate enforcement into a cohesive workflow.

This command operates under the L2+ authority level and is executed by the cicd-tooling-specialist agent. It is part of the platform's 216-command slash command registry, built on the AIAD (Autonomous Intelligence Agent Design) standard. The unified approach ensures that all CI/CD operations are subject to the same quality enforcement, telemetry tracking, and doctrine compliance as every other platform command.

The Prismatic Platform's CI/CD infrastructure manages a complex pipeline that must compile nearly 100 umbrella applications, run thousands of tests, enforce zero-warning compilation, execute Credo static analysis, verify Dialyzer type specifications, and deploy to multiple environments. The /cicd-unified command tames this complexity by providing intelligent defaults, automatic pipeline stage detection, and integration with the platform's quality gate system.

Architecture

The command is architected as a facade over the platform's CI/CD subsystem, which itself integrates with GitLab CI, Fly.io deployment, and the internal quality gate infrastructure.

Pipeline Model

The platform uses a multi-stage pipeline model defined in .gitlab-ci.yml. Each stage corresponds to a quality level that must be satisfied before proceeding.

Stage	Purpose	Gate
compile	Zero-warning compilation with `--warnings-as-errors`	HARD BLOCK
lint	Credo strict analysis and formatting checks	HARD BLOCK
typecheck	Dialyzer type specification verification	HARD BLOCK
test	Full test suite execution with coverage	HARD BLOCK
quality	Quality gate composite score verification	HARD BLOCK
build	Docker image construction (multi-stage)	HARD BLOCK
deploy-staging	Fly.io staging environment deployment	SOFT (manual trigger)
deploy-production	Fly.io production environment deployment	SOFT (manual trigger)

Component Topology

/cicd-unified
    |
    +-- PipelineController
    |     |-- GitLabClient (REST API)
    |     |-- PipelineMonitor (polling/webhook)
    |     +-- ArtifactManager (download/cache)
    |
    +-- DeploymentController
    |     |-- FlyClient (fly.io CLI wrapper)
    |     |-- HealthChecker (post-deploy verification)
    |     +-- RollbackManager (automatic rollback)
    |
    +-- QualityGateEnforcer
          |-- CompilationChecker
          |-- CredoRunner
          |-- DialyzerRunner
          +-- TestRunner

Usage

# Trigger a new pipeline for the current branch
/cicd-unified trigger

# Check the status of the current pipeline
/cicd-unified status

# View detailed pipeline logs for a specific stage
/cicd-unified logs --stage=test

# Deploy to staging after pipeline passes
/cicd-unified deploy --target=staging

# Deploy to production with approval workflow
/cicd-unified deploy --target=production --approve

# Rollback production to previous release
/cicd-unified rollback --target=production

# List recent pipeline runs with results
/cicd-unified history --limit=10

# Download build artifacts
/cicd-unified artifacts --pipeline=latest --stage=build

# Retry a failed pipeline stage
/cicd-unified retry --stage=test --pipeline=latest

# Run quality gates locally before triggering remote pipeline
/cicd-unified preflight

Options

Option	Type	Default	Description
`action`	string	required	Action: trigger, status, logs, deploy, rollback, history, artifacts, retry, preflight
`--target`	string	staging	Deployment target: staging, production
`--stage`	string	all	Pipeline stage to operate on
`--pipeline`	string	latest	Pipeline ID or "latest"
`--branch`	string	current	Git branch for pipeline trigger
`--approve`	flag	false	Require explicit approval for deployment
`--limit`	integer	5	Number of history entries to display
`--format`	string	text	Output format: text, json, table
`--wait`	flag	false	Block until pipeline completes
`--timeout`	integer	1800	Timeout for `--wait` in seconds
`--verbose`	flag	false	Show detailed output including job logs
`--dry-run`	flag	false	Show what would happen without executing

Execution Flow

The execution flow varies by action, but all actions share a common preamble of authentication, context resolution, and quality pre-checks.

Authentication: GitLab API token and Fly.io credentials are validated from environment variables (GITLAB_TOKEN, FLY_API_TOKEN).
Context Resolution: The current git branch, latest commit SHA, and pipeline state are resolved to establish the operational context.
Pre-Check: Quality gates are verified locally to catch obvious issues before triggering remote operations. This prevents wasting CI minutes on pipelines that will fail in the first stage.
Action Dispatch: The requested action is dispatched to the appropriate controller (PipelineController for trigger/status/logs/history/retry, DeploymentController for deploy/rollback).
Execution: The action executes with full telemetry instrumentation. Pipeline triggers use the GitLab REST API, deployments use the Fly.io CLI, and quality checks use local mix tasks.
Result Reporting: Results are formatted and presented with clear status indicators. Failed operations include diagnostic information and suggested remediation steps.
Post-Action Hooks: Session lifecycle hooks record the CI/CD action for audit trail purposes. Deployment actions additionally trigger health checks on the target environment.

Integration Points

Component	Relationship	Details
Prismatic Agents	Executed by `cicd-tooling-specialist`	Agent manages pipeline lifecycle
GitLab CI	Pipeline backend	REST API for trigger, status, logs, artifacts
Quality Gates	Pre/post execution validation	Local preflight and remote stage gates
Telemetry	Execution metrics	Pipeline duration, stage timing, deployment frequency
AIAD Registry	Command specification	CI/CD command configuration
Fly.io	Deployment target	Staging and production hosting
Docker	Image building	Multi-stage builds for production
/guardrails	Deployment safety	Pre-deployment safety checks

Best Practices

Run preflight before triggering. The preflight action runs all quality checks locally in under 60 seconds. This catches compilation warnings, Credo violations, and test failures before consuming CI minutes on the remote pipeline.

Use staging before production. Always deploy to staging first and verify the application health before promoting to production. The command enforces this by requiring --approve for production deployments.

Monitor pipeline duration trends. Use history to track pipeline duration over time. A gradually increasing pipeline time often indicates growing test suites or dependency resolution issues that should be addressed.

Pin pipeline references. When discussing specific pipelines in issues or documentation, use the pipeline ID rather than "latest" to ensure the reference remains stable over time.

Configure timeout appropriately. The full pipeline for the Prismatic Platform with nearly 100 umbrella apps takes 15-25 minutes. Set --timeout accordingly when using --wait.

Use atomic deployments. The Fly.io deployment strategy uses blue-green deployment with health checks. If the new version fails health checks, the deployment automatically rolls back without operator intervention.

Error Handling

Error	Cause	Resolution
`gitlab_auth_failure`	Invalid or expired `GITLAB_TOKEN`	Regenerate token in GitLab settings
`pipeline_trigger_failed`	Branch protection or CI configuration error	Verify branch permissions and `.gitlab-ci.yml`
`stage_timeout`	Pipeline stage exceeded time limit	Check for hanging tests or resource contention
`deploy_health_check_failed`	Application failed post-deploy health check	Automatic rollback triggered, investigate logs
`fly_auth_failure`	Invalid `FLY_API_TOKEN`	Re-authenticate with `fly auth token`
`quality_gate_blocked`	Pre-flight quality check failed	Fix reported issues before triggering pipeline
`artifact_not_found`	Requested artifact expired or never created	Check pipeline configuration for artifact retention
`rollback_failed`	No previous release available	Manual intervention required via Fly.io dashboard

Advanced Usage

Pipeline Scheduling

Schedule pipelines for off-peak execution to optimize CI resource utilization.

# Schedule pipeline for 2 AM
/cicd-unified trigger --schedule="02:00" --branch=main

# Trigger nightly full pipeline
/cicd-unified trigger --full --schedule="nightly"

Custom Pipeline Variables

Pass custom variables to pipeline runs for conditional stage execution.

# Skip deployment stages
/cicd-unified trigger --var="SKIP_DEPLOY=true"

# Run extended test suite
/cicd-unified trigger --var="TEST_SUITE=extended"

Multi-Environment Deployment

Deploy specific commits to specific environments with environment-specific configuration.

# Deploy specific SHA to staging
/cicd-unified deploy --target=staging --sha=abc1234

# Promote staging to production (uses staging's verified image)
/cicd-unified deploy --target=production --promote-from=staging

GitLab CI YAML Compliance

The platform enforces strict GitLab CI YAML patterns. The command validates YAML before submission.

# Validate CI configuration
/cicd-unified validate-yaml

# Lint CI configuration with platform standards
/cicd-unified lint-yaml --strict

Doctrine Compliance

All commands operate under the NO MERCY, NO DOUBTS doctrine:

NO MERCY: Zero tolerance for incomplete execution or quality violations. Every pipeline stage must pass before deployment is permitted. No bypass flags.
NO DOUBTS: Full investigation before action, evidence-based results. Pipeline results are fully auditable with complete logs and artifact preservation.

/guardrails - CI/CD guardrails enforcement for deployment safety
/gitlab-ci - GitLab CI/CD pipeline management and configuration
/gitlab-api - GitLab API operations for project and repository management
/quality-gates - Enforce quality gate checkpoints with zero-warning compilation validation
/commit - Smart commit with quality gates and conventional format
/connect - MCP server connection management across 14+ servers

Connect & Contribute

Created by Tomáš Korcak (korczis) | Open Source under GHL

GitHub | GitLab | LinkedIn | Contact
Developer Portal | Architecture | Meet the Creator

/cicd-unified