# Assay

> Assay is the quality layer for agentic software. It rates MCP servers, APIs, and SDKs across agent friendliness, security, and reliability — so agents and developers can choose the right tools.

Assay evaluates packages on three dimensions (each 0-100):
- **Agent Friendliness (AF)** — MCP quality, docs, error messages, auth simplicity, rate limits
- **Security** — TLS, auth strength, scope granularity, dependency hygiene, secret handling
- **Reliability** — uptime, version stability, breaking changes, error recovery

## API

- [Package list](/v1/packages): Browse and filter evaluated packages.
- [Package detail](/v1/packages/{package_id}): Full evaluation data for a single package.
- [Agent guide](/v1/packages/{id}/agent-guide): Agent-optimized view with scores.
- [Categories](/v1/categories): List all categories with package counts.
- [Category packages](/v1/categories/{slug}/packages): Packages in a category, by AF score.
- [Compare](/v1/compare?ids=a,b,c): Side-by-side comparison of up to 10 packages.
- [Stats](/v1/stats): Sitewide statistics and score distribution.
- [Evaluation queue](/v1/queue): Packages needing evaluation or re-evaluation.
- [Health](/v1/health): Health check endpoint.

## Website

- [Homepage](https://assay.tools/): Browse top-rated packages and categories.
- [All packages](https://assay.tools/packages): Search and filter the full directory.
- [Categories](https://assay.tools/categories): Browse by category.
- [Compare](https://assay.tools/compare): Side-by-side package comparison.
- [Contribute](https://assay.tools/contribute): See the evaluation queue and help rate packages.
- [Evaluation Guide](https://assay.tools/evaluate.md): Complete rubric and submission instructions.
- [About](https://assay.tools/about): Scoring methodology and coverage stats.

## Optional

- [OpenAPI spec](/openapi.json): Full API schema in OpenAPI 3.1 format.
- [MCP server](https://github.com/Assay-Tools/assay): Assay's MCP server for agent integration.