Skip to main content
ValidationForge · Skills · 52 Skills · 8 Categories

52 Skills.
Every validation need covered.

Skills are the execution units of ValidationForge. Each skill is a focused, composable capability — a platform runner, a quality gate, a synthesis protocol. Commands invoke skills; you configure enforcement levels.

ValidationForgeSkills

52
Total Skills
15
Platform
6
Quality Gates
7
Specialized
3
Consensus
ios-validation
Full iOS simulator journey execution via xcrun simctl + idb. Screenshots + a11y tree at every transition.
ios-validation-gate
Hard gate: blocks PASS until all iOS PASS criteria have non-zero evidence files.
ios-validation-runner
Step executor for iOS journeys. Handles boot, deeplink, tap, scroll, assert.
ios-simulator-control
Low-level xcrun simctl wrapper: boot, install, launch, screenshot, log capture.
playwright-validation
Playwright MCP integration for web journeys. Headless or headed, captures DOM snapshots + screenshots + network logs.
web-validation
High-level web journey runner. Wraps playwright-validation with VF evidence discipline.
web-testing
Interaction testing for web: form fill, navigation, state assertions, network intercept.
chrome-devtools
Deep browser debugging via Chrome DevTools MCP. Memory snapshots, performance traces, console capture.
api-validation
HTTP endpoint validation via curl. Asserts status codes, response bodies, headers, latency.
cli-validation
Direct CLI invocation validation. Captures stdout/stderr, asserts exit codes, timeout handling.
fullstack-validation
Multi-layer validation for fullstack projects (frontend + backend + DB). Coordinates per-layer validators.
react-native-validation
React Native app validation via Detox/Appium. Device screenshots + interaction logs for iOS + Android.
flutter-validation
Flutter app validation via Flutter Driver + integration_test. Widget tree dumps + screenshots.
django-validation
Django/Flask API validation. runserver + curl + log capture. Asserts routes, responses, middleware.
rust-cli-validation
Rust CLI validation. cargo run + binary execution. stdout/stderr capture + exit code assertions.
functional-validation
Core no-mock mandate enforcer. Validates that the real system was exercised, not a mock or stub.
gate-validation-discipline
Gate checker: verifies every PASS verdict has cited evidence before allowing pipeline to advance.
no-mocking-validation-gates
Blocks test files, mocks, stubs, test doubles at write time. Iron Rule 2 enforcement.
build-quality-gates
Preflight build gate. Fails the run if the project does not compile cleanly.
verification-before-completion
Completion claim validator. Ensures 'done' is backed by evidence, not just build success.
preflight
System readiness check. Verifies server is running, MCP servers are available, build is clean.
design-validation
Design fidelity audit. Compares rendered UI against design tokens and visual spec.
design-token-audit
Checks that deployed CSS variables match the design system token definitions (colors, fonts, radii).
stitch-integration
Stitch MCP integration for design screenshot generation and comparison.
visual-inspection
Screenshot-based UI defect classification. Uses vision model to flag layout, color, and typography issues.
sequential-analysis
Multi-step root cause investigation for FAIL verdicts. Traces evidence chain to source code.
research-validation
Standards research: WCAG, HIG, OWASP. Maps applicable criteria to VF journeys.
retrospective-validation
Post-run retro: what failed, why, what the fix was, what to add to the journey next time.
ai-evidence-analysis
Vision model pre-screening of captured screenshots. Flags obvious failures before human review.
accessibility-audit
WCAG 2.1 AA compliance audit. Runs axe-core assertions against live rendered pages.
responsive-validation
Multi-viewport layout validation: 375px / 768px / 1440px. Screenshots at each breakpoint.
parallel-validation
Orchestrates parallel journey execution across multiple sub-agents. Collects results for unified verdict.
coordinated-validation
Dependency-aware multi-platform coordination. Blocks downstream validators until upstream PASS.
e2e-testing
End-to-end journey execution framework. Wraps platform validators with evidence discipline.
e2e-validate
Cross-domain navigation validation: post → catalog → product site → GitHub. Verifies link graph.
create-validation-plan
Journey plan generator. Scans codebase, produces YAML with PASS criteria and evidence specs.
baseline-quality-assessment
Establishes a validation baseline for a project with no prior VF history.
condition-based-waiting
Waits for async conditions (server ready, animation complete, API response) before capturing evidence.
error-recovery
Handles transient errors during sweep: network timeouts, flaky selectors, slow renders.
production-readiness-audit
Pre-deploy checklist: security headers, error pages, env vars, health endpoint, HTTPS.
full-functional-audit
Comprehensive single-pass audit: all journeys, all platforms, all evidence. No partial runs.
evidence-dashboard
Generates evidence summary dashboard (markdown + HTML) from captured e2e-evidence. Per-journey verdicts, quality scores, historical trends — the artifact engineering leads review for ship / no-ship.
forge-setup
Initialize FORGE engine. Writes forge-config.json and creates the runs/ archive directory.
forge-plan
Generate a forge execution plan. Supports --consensus for 3-perspective synthesis.
forge-execute
Execute forge pipeline with hard phase gates. State persists to forge-state.json.
forge-team
Multi-agent forge execution with dependency-aware validator waves.
forge-benchmark
Benchmark FORGE runs: fix efficiency, attempt distribution, UNFIXABLE rate.
validate-audit-benchmarks
Cross-run audit of benchmark history. Flags regressions, highlights trends.
team-validation-dashboard
Dashboard generator for multi-agent team validation posture.
consensus-engine
Coordinator for multi-validator consensus runs. Spawns N validators, monitors completion, invokes synthesizer.
consensus-synthesis
Synthesizer: reads all N validator verdicts, applies synthesis-states table, emits confidence-scored verdict.
consensus-disagreement-analysis
Disagreement protocol: classifies diverging criteria as missing evidence / contradictory / interpretation / validator error.
bashskill invocation patterns
# Commands invoke skills automatically based on platform detection
/validate                  # → invokes web-validation or ios-validation etc.
/validate-sweep            # → preflight + platform validator + verdict-writer
/validate-consensus        # → consensus-engine + N × functional-validation + consensus-synthesis

# Skills also activate on trigger keywords (in CLAUDE.md)
# "validate this form"   → functional-validation
# "check accessibility"  → accessibility-audit
# "run sweep"            → validate-sweep → web-validation

# Invoke directly (advanced)
/oh-my-claudecode:skill validationforge:ios-validation
/oh-my-claudecode:skill validationforge:consensus-engine