Author and validate Agent QA test definitions with schema compliance

Agent QA Authoring is a reusable skill that enables AI assistants to create, edit, validate, and execute Agent QA tests, suites, and hooks without inventing schema fields or identifiers. It enforces strict ID contracts (test IDs start with t_, suite IDs with s_, hook IDs with h_) and validates every definition against the Agent QA schema before saving or running. The skill prefers MCP tools for authoring mutations and falls back to CLI or YAML when MCP is unavailable.

Use this skill when creating or editing an Agent QA test, suite, or hook; validating Agent QA YAML or canonical IDs; running a newly authored definition through MCP or CLI; or investigating which configuration fields or workspace patterns apply. Do NOT use this skill to invent undocumented config keys, hand-write IDs, mutate files outside configured workspace patterns, or run destructive or production-facing scenarios without explicit user confirmation. The skill does not infer undocumented config keys, selectors, UI states, credentials, or test data.

You provide a configured Agent QA workspace, authorization scope, and the target environment (staging, production, etc.). The skill discovers the local surface with agent_qa_discover, inspects active config with agent_qa_get_config, and loads references/agent-qa-contracts.json for exact schema fields. It generates every new ID using agent_qa_generate_id (MCP) or agent-qa ids generate <test|suite|hook|run|observation> (CLI), validates definitions with agent_qa_validate_test, agent_qa_validate_suite, or agent_qa_validate_definition, and creates or updates definitions through MCP authoring mutations (agent_qa_create_test, agent_qa_update_test, agent_qa_create_suite, etc.).

Agent QA Authoring works with Claude, Cursor, Gemini, and Codex. It supports web-testing and mobile-testing workflows and integrates with any browser, mobile, model-provider, or application dependencies configured in the Agent QA workspace target.

This skill is for QA engineers, test automation developers, and AI assistant users who need to author Agent QA tests programmatically while maintaining strict schema compliance. It is critical-risk because it can trigger test runs that mutate real data or external application state. The skill requires an installed and configured Agent QA workspace plus dependencies for the selected target. Here is a concrete example of the expected workflow:

User: Add an Agent QA checkout test for the staging target and validate it, but do not run it yet.

Expected handling: discover the workspace, inspect the staging target, generate the test ID,
create the smallest valid definition, validate it, and stop before enqueueing a run.

The skill enforces required ID contracts: test IDs use t_ plus 10 id-agent words, suite IDs use s_ plus 10 id-agent words, hook IDs use h_ plus 10 id-agent words, run IDs use r_ plus 10 id-agent words, and observation IDs use obs_ plus 10 id-agent words. Validation proves schema compatibility but does not guarantee that the application behavior or external environment is safe to exercise.