> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/docsfab/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/docsfab/_mcp/server.

# Assessment and Evaluation Skills in DocsFab

> Look up the DocsFab skill contracts for documentation-set audits, audience testing, and quality evaluation.

These DocsFab skills assess a frozen target without rewriting it. They
preserve missing evidence, incompatible findings, and population
differences instead of averaging them into an unsupported pass. For more
information, refer to [Skill Behavior Contracts in DocsFab](/docsfab/reference/quality/skills).

## Skill Contracts

| Skill | Primary question | Result boundary |
| --- | --- | --- |
| [Audit Documentation Skill in DocsFab](/docsfab/reference/quality/skills/assess-and-evaluate/audit-documentation) | What is in scope, supported, drifting, missing, or unassessed? | Inventory and findings; no release verdict. |
| [Test Audience Understanding Skill in DocsFab](/docsfab/reference/quality/skills/assess-and-evaluate/test-audience-understanding) | What does a representative human or clean agent understand from the frozen documentation? | Controlled population-specific evidence; no coaching during measurement. |
| [Evaluate Document Quality Skill in DocsFab](/docsfab/reference/quality/skills/assess-and-evaluate/evaluate-document-quality) | What does the complete layered evidence package establish? | Hard-gate-aware quality result; missing required evidence cannot be averaged away. |