Connect Any AI Agent to Glintbase ARS
Zero-config agent readiness tools for Claude Desktop, Cursor, ChatGPT, and Windsurf. Run live audits, multi-modal flight simulations, and in-memory sandboxes directly from your AI pair programmer.
Streamable Remote MCP Connector
Connect directly via public HTTPS URL without installing packages or configuring API keys.
{
"mcpServers": {
"glintbase": {
"url": "https://scan.glintbase.dev/api/mcp"
}
}
}Paste into %APPDATA%\Claude\claude_desktop_config.json (Windows) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS).
The 17 Production MCP Tools
Engineered for low token consumption, high signal density, and multi-modal feedback.
Complete ARS 3.0 audit across 6 pillars
Evaluates Discovery, Access, Usability, Semantic, Architecture, and Safety with dynamic denominator scoring.
Ultra-compact ARS 3.0 scorecard
Returns letter grade, score (0-100), and gate status without burning the agent's context window.
Discovers all machine entrypoints
Checks robots.txt, llms.txt, .well-known/ard.json, auth.md, OpenAPI, and WebMCP routes.
Multi-agent flight simulator with visual graph
Simulates Claude Code, Cursor, or Perplexity navigating your API. Returns JSON telemetry, a dark-mode Journey Tree SVG image, and an instant replay link.
Before/after empirical proof card
Compares baseline performance against performance with living artifacts installed, proving token savings and failure reduction.
Quantifies context window bloat
Calculates the token tax multiplier incurred by unstructured HTML documentation vs machine-readable specs.
Evaluates hallucination risk on parameters
Inspects API and MCP parameter schemas for ambiguity, excessive nesting, and type friction.
Synthesizes living artifacts with AST validation
Generates production-grade llms.txt, ard.json, auth.md, or streamable MCP route handlers.
In-memory AST evaluation without disk writes
Tests proposed changes and artifact structures in an isolated in-memory sandbox before applying them.
Client-side WebMCP window.modelContext audit
Audits frontend interfaces for semantic agent forms, microdata, and client-side tool registrations.
CI quality gate with PR comments
Evaluates git drift against an ARS threshold and outputs ready-to-paste GitHub step summary markdown.
OWASP Top 10 for LLMs & ISO 42001 assessment
Produces executive-level compliance mapping across prompt injection barriers, data boundary leakage, and agent auth.
Detects soft-200 SPA leaks & 404 traps
Probes synthetic 404 routes to ensure the API returns strict RFC 7807/9457 error bodies instead of HTML pages.
Validates WorkOS auth.md & RFC 9728 credentials
Verifies machine authentication surfaces, OAuth2 client credential discovery, and scoped permissions.
Audits Idempotency-Key & mutation hints
Checks state-changing POST/PUT/DELETE endpoints for idempotency locks, readOnlyHint, and destructiveHint flags.
Retrieves complete markdown skill instructions
Loads any of the 9 bundled Glintbase skills as programmatic agent context.
Scaffolds .agents/skills/<name>/SKILL.md to disk
Directly writes the chosen skill playbook into the project repository for permanent agent pair-programming.
9 Bundled Agent Skills
Registered as native skill://glintbase/* resources and optimization prompts.
npx @glintbase/cli skill install <name>Glintbase Agent Readiness (ARS 3.0)
The master ARS 3.0 specification covering 6 pillars, dynamic denominator scoring, and automated drift prevention.
Living Artifacts Architect
Best practices for authoring public/llms.txt, llms-full.txt, and .well-known/ard.json.
Streamable MCP Builder
Guide to building Model Context Protocol servers with dual Stdio & Streamable HTTP transports.
Agent Authentication Handbook
Machine-to-machine authentication with RFC 9728, WorkOS auth.md, and scoped API key handshakes.
WebMCP Browser Integration
Registering client-side tool capabilities on window.modelContext for browser-based agents.
Token Tax & Schema Optimizer
Techniques to eliminate context window bloat, reduce parameter cognitive friction, and avoid hallucinations.
Flight Simulator Replay
Debug agent navigation trajectories, analyze breadcrumb traces, and prove counterfactual fixes.
Zero-Drift CI Gate
Automating ARS regression prevention in GitHub Actions with step summaries and PR comments.
Enterprise Agent Governance
Mapping agent readiness to OWASP Top 10 for LLMs, ISO/IEC 42001, and SOC2 audit compliance.
Watch Agents Navigate APIs Live in the Web Cockpit
Replay simulation runs generated by MCP, examine Journey Trees, and test counterfactual What-If fixes in real-time.