# Glintbase — Agent Readiness Infrastructure
> Glintbase helps companies measure, understand, and improve how AI agents discover, navigate, and operate software across documentation, APIs, onboarding, SDKs, and support surfaces.

## Core Machine Surfaces & Plain-Text Endpoints
- [/llms.txt](https://glintbase.dev/llms.txt): Plain-text machine index (this file).
- [/llms-full.txt](https://glintbase.dev/llms-full.txt): Comprehensive concatenated documentation & research manuscript for LLM ingestion.
- [/research.md](https://glintbase.dev/research.md): Full 24-page research manuscript: "State of Agent Readiness 2026".
- [/docs.md](https://glintbase.dev/docs.md): Developer documentation in raw Markdown format (CLI, MCP Server, Agent Skill, ARS 1.0).
- [/blog.md](https://glintbase.dev/blog.md): Catalog of all Glintbase research journal articles in raw Markdown format.
- [/mcp.json](https://glintbase.dev/mcp.json): Model Context Protocol server manifest (@glintbase/mcp).
- [/reports/STATE_OF_AGENT_READINESS_2026.pdf](https://glintbase.dev/reports/STATE_OF_AGENT_READINESS_2026.pdf): Published 24-page research PDF (1.1 MB).
- [/research/data/full-scores.csv](https://glintbase.dev/research/data/full-scores.csv): Raw 100-platform cohort benchmark scores CSV.
- [/research/data/research-dataset.json](https://glintbase.dev/research/data/research-dataset.json): Full 100-platform empirical dataset JSON.

## Key Research & Benchmark Publications
- [State of Agent Readiness 2026](https://glintbase.dev/research): Global Benchmark Report evaluating 100 AI engineering platforms across 1,000 Pathfinder scans and 450 live coding agent executions. Mean ARS Score: 50.7/100 (Grade C).
- [The Agent Resilience Paradox](https://glintbase.dev/research.md#63-the-agent-resilience-paradox-in-detail): Why smart LLMs succeed despite bad docs by paying a massive token tax ($433.2k/yr per team).
- [9-Action Remediation Playbook](https://glintbase.dev/research.md#remediation-playbook): Actionable steps to boost platform ARS scores by up to +35 points.

## Developer Products & Tools
- [Glintbase Hosted Scanner](https://scan.glintbase.dev): Probe any documentation or product URL to calculate its Agent Readiness Score (0-100).
- [Glintbase CLI (@glintbase/cli)](https://glintbase.dev/docs#cli): Terminal scanner and CI/CD quality gate for agent readiness.
- [Glintbase MCP Server (@glintbase/mcp)](https://glintbase.dev/docs#mcp): Model Context Protocol server with 9 readiness tools.
- [Glintbase Agent Skill](https://glintbase.dev/docs#skill): Installable skill teaching coding agents to scan and fix repos autonomously.

## Research Journal Articles (Raw Markdown)
- [Agent Readiness: The Missing Metric in Modern Software](https://glintbase.dev/blog/Agent-readiness-missing-metric-in-modern-software)
- [Why AI Agents Need a Different Interface Than Humans](https://glintbase.dev/blog/AI-agents-needs-a-different-interface)
- [Why Most RAG Fails for Technical Documentation](https://glintbase.dev/blog/why-most-rag-fails)
- [Documentation Drift is Killing AI Agents](https://glintbase.dev/blog/docs-drift-is-killing-ai)
- [Why Agents Use the Web Differently Than Browsers](https://glintbase.dev/blog/why-agents-use-the-web-differently)
- [The Future of Agent-Operable Software](https://glintbase.dev/blog/future-of-agent-operable-software)
- [What is Agent Readiness?](https://glintbase.dev/blog/what-is-ai-readiness)
- [The $433k Token Tax: Why Bad Docs Drain AI Infrastructure Budgets](https://glintbase.dev/blog/token-tax-ai-docs)
- [The Agent Resilience Paradox: Why 'Working' Software Fails AI Agents](https://glintbase.dev/blog/agent-resilience-paradox)
- [Beyond /llms.txt: The Full Machine-Readable Stack for AI Agents](https://glintbase.dev/blog/llms-txt-machine-surfaces)
- [State of Agent Readiness 2026: What We Found Across 75 Software Platforms](https://glintbase.dev/blog/state-of-agent-readiness-2026)
- [How to Gate CI/CD on Agent Readiness: A Practical Glintbase CLI Tutorial](https://glintbase.dev/blog/agent-readiness-cicd-gate)

## Open Source Repositories
- [glintbase/ars-report](https://github.com/glintbase/ars-report): Research report build engine & benchmark execution suite.
- [glintbase/glintscanner](https://github.com/glintbase/glintscanner): Core scanner engine, CLI, and MCP server.