BlankVisuals© Validator

Evidence-backed validation framework for AI agents, Cursor skills, MCP orchestrations, subagent graphs, hooks, and agentic repos — before deployment, not after.

Field	Value
Framework ID	FW-001
Full name	BlankVisuals© Validator
Version	v1.0.2
Date registered	2026-05-20
Domain package	cursor-skills
Method	precise & verify — every rule needs verified evidence or INCONCLUSIVE
License	MIT

What it does

BlankVisuals© Validator is an open-source AI agent validation framework built as a Cursor Agent Skill.

When invoked, the agent:

1. Discovers your target (repo or agent built on FW-001)
2. Resolves validation rules (your spec → repo VALIDATION.md → built-in catalog)
3. Runs checks in parallel where safe, then dependent checks (topology, E2E)
4. Records verified evidence per rule (commands, file citations, tool results)
5. Delivers a validation canvas plus a short chat summary

It does not guess pass/fail. Unverified items are marked INCONCLUSIVE, not PASS.

Installation

Option A — Personal skill (recommended)

Clone or copy this folder into your Cursor skills directory:

git clone https://github.com/cengicadis/blankvisuals-validator ~/.cursor/skills/blankvisuals-validator

Cursor loads skills from ~/.cursor/skills/ automatically when the skill description matches your request.

Option B — Project skill

Copy into a project:

your-project/.cursor/skills/blankvisuals-validator/

Commit SKILL.md, reference.md, and .gitignore with your project if the team should share the same validation workflow.

Do not commit

• .vscode/launch.json and .vscode/*.code-workspace — local only (ignored via .gitignore)
• .DS_Store — macOS metadata (ignored)

Committed with project defaults

• .vscode/settings.json — project-only keys; no personal editor prefs (v1.0.2+)

Quick start

1. Open the repo or agent you want to validate in Cursor.
2. In chat, ask for validation, for example:
   • "Run BlankVisuals validator on this repo"
   • "Validate this agent against FW-001"
   • "Validate this MCP orchestration against FW-001"
   • /validate this orchestration workflow
3. Review the canvas (canvases/blankvisuals-validation-<target>.canvas.tsx) and the chat summary.

The agent stamps framework meta (FW-001) at the start and sets validation confirmation only after evidence is complete.

Documentation

Document	Purpose
docs/USAGE.md	Step-by-step usage, prompts, deliverables, custom rules
SKILL.md	Agent orchestration workflow (for the skill itself)
reference.md	Full rule catalog (STR, TOP, CON, SEC, RUN, …)

Wiki

Full documentation available on the GitHub Wiki:

Page	What you'll find
Getting Started	Install, first run, example prompts
How It Works	Architecture, six domains, evidence standard
Rule Catalog	Every rule with check, pass, and fail conditions
Framework Types	How to classify your target
Custom Rules	VALIDATION.md format and examples
Troubleshooting	Common issues and fixes

Repository layout

blankvisuals-validator/
├── SKILL.md          # Cursor skill — orchestration & evidence standards
├── reference.md      # FW-001 rule catalog & taxonomy (canonical)
├── VALIDATION.md     # Repo verify spec (RUN-01)
├── scripts/
│   └── verify.sh     # Structure verify command
├── docs/
│   └── USAGE.md      # Human-facing usage guide
├── .gitignore
├── LICENSE
└── README.md         # This file

Canonical rules: reference.md. Verify: bash scripts/verify.sh (see VALIDATION.md).

Targets in scope

Validate any AI agent, prompt workflow, or agentic repo against a structured rule catalog with verified evidence per finding.

Supported targets:

• Skill chains (SKILL.md, skill frontmatter)
• Hook pipelines (hooks.json)
• Subagent graphs (Task tool, resume/interrupt)
• MCP orchestration
• @cursor/sdk automation
• Plugin marketplace layouts (.cursor-plugin, manifests)

See Framework taxonomy in reference.md.

Custom rules for your repo

Add a repo-level spec so validation matches your project:

• VALIDATION.md (repo root), or
• docs/validation.md, or
• .cursor/validation/

Format and examples: docs/USAGE.md#custom-validation-rules.

Outputs

Deliverable	Location / format
Validation canvas	canvases/blankvisuals-validation-<slug>.canvas.tsx
Chat summary	Markdown in chat (FW-001 meta, pass/fail counts, critical IDs)
Evidence	Per-rule YAML blocks (command / file / tool)

Anti-patterns (avoid)

• Claiming PASS without command output or file citations
• Setting validation_confirmation before checks finish
• Using markdown tables instead of a canvas for the main report
• Validating ~/.cursor/skills-cursor/ (reserved)

Versioning

Bump the patch version on each release (1.0.0 → 1.0.1 → 1.0.2 → …) in SKILL.md, reference.md, README.md, docs/USAGE.md, and VALIDATION.md. See CHANGELOG.md.

License

MIT — see LICENSE. Free to use, modify, and distribute with attribution.

Attribution

BlankVisuals© Validator · Framework FW-001 · registered 2026-05-20.