Marsley Mash
Community@marsley01 · Nairobi
breaking and learning
Agent Skills by Marsley Mash
Showing 33 vetted skills indexed across 1 GitHub repositories.
marsley01vc-audit-context
Validates context routing, skill discoverability, and Claude/Codex wiring with Node.js validator scripts.
marsley01vc-validate-findings
Orchestrates two-layer parallel validation of implementation plans into a PASS, CONDITIONAL, or BLOCKED gate verdict.
marsley01vc-generate-phase-program
Generate kickoff artifacts for multi-phase RIPER-5 programs including umbrella plans and phase stubs.
marsley01vc-debug
Diagnose bugs and system failures through root-cause investigation before applying fixes.
marsley01vc-scenario
Generates edge cases and test scenarios by decomposing features across 12 risk dimensions.
marsley01vc-update
Syncs agent harness files from a remote kit repository with dry-run diff and confirmation.
marsley01vc-agent-browser
Automates browser interactions via CLI using context-efficient accessibility snapshots with element refs.
marsley01vc-autoresearch
Runs iterative find-gaps, fix, and re-verify loops until convergence or a metric target is reached.
marsley01vc-context-discovery
Discovers and loads repository context files by frontmatter routing and keyword matching.
marsley01vc-frontend-design
Create production-grade frontend interfaces from screenshots, videos, or design briefs with anti-slop aesthetic rules.
marsley01vc-plan-discovery
Discovers and groups plan artifacts by scanning frontmatter across feature folders.
marsley01vc-audit-vc
Validates agent harness consistency across Claude, Codex, README, and protocol files.
marsley01vc-generate-spec
Generates plain-language product-discovery SPEC requirements documents from research findings and user intent.
marsley01vc-generate-plan
Generates structured implementation plan artifacts in SIMPLE or COMPLEX formats with validation.
marsley01vc-sequential-thinking
Applies structured step-by-step reasoning with revision, branching, and hypothesis verification to complex problems.
marsley01vc-risk-evidence-pack
Generate and validate five-artifact evidence packs for high-risk code changes before finalization.
marsley01vc-generate-closeout
Generates post-execution closeout packets with drift scoring and archive-readiness classification for completed plans.
marsley01vc-predict
Simulates five expert personas debating proposed changes to surface architectural, security, and performance risks before implementation.
marsley01vc-generate-context
Generates and updates the authoritative repository context file at process/context/all-context.md.
marsley01vc-publish
Publishes harness improvements from a development repo to the remote kit repository with diffing, versioning, and leak detection.
marsley01vc-scout
Searches codebases with shell tools and parallel agents to locate relevant files.
marsley01vc-security
Audits code for STRIDE and OWASP vulnerabilities with optional iterative auto-fix.
marsley01vc-feasibility-test
Runs empirical probes against unverified runtime or library mechanisms and produces a structured verdict artifact.
marsley01vc-test-coverage-plan
Generates tiered test coverage plans assigning four test tiers per blast radius area.
Frequently Asked Questions About Marsley Mash
FAQPage SchemaWhat tasks can I accomplish with Marsley Mash's skill kit?▼
You can audit context routing and skill discoverability, generate specs and implementation plans, run validation fan-outs with net gates, debug with root-cause analysis, perform STRIDE/OWASP security audits, run Playwright/Vitest/k6 web tests, and manage plan closeouts with drift scoring.
Who is the target user for these skills?▼
Developers running Claude or Codex agent sessions who need structured harness governance: context discovery, plan artifacts, feasibility probes, risk evidence packs, and test coverage planning across multi-phase programs.
How do I install and start using the harness?▼
Run vc-setup for interactive harness setup: it detects your stack, scaffolds process directories, deep-scans the codebase, and populates context. Use vc-update to pull kit improvements and vc-publish to push changes back to the remote kit repository.
What licensing applies to these skills?▼
Licensing varies per skill: vc-scenario, vc-sequential-thinking, vc-predict, vc-security, and vc-review-situation are MIT; vc-agent-browser and vc-web-testing are Apache-2.0; vc-frontend-design references terms in LICENSE.txt. Several patterns credit autoresearch by Udit Goenka.
What prerequisites or dependencies do the skills require?▼
Skills operate inside a Claude/Codex agent harness with a process/context directory structure and plan artifacts. Web testing requires Playwright, Vitest, or k6; documentation lookup uses llms.txt via context7.com; cloud browser sessions optionally use Browserbase.