Skills reference

The repository ships three Agent Skills under skills/. Each is an agentskills.io-compliant package (SKILL.md plus progressive-disclosure references/ and, where applicable, scripts/ and assets/) that activates when a prompt matches its trigger keywords.

Skill	Purpose	Category	Needs a live server?
a2acli	Drive the a2acli CLI to discover, invoke, and monitor A2A agents	cli-usage	Yes
a2a-expose	Add an A2A v1.0 exposure layer to an existing API or service	protocol-engineering	No (builds one)
a2a-conformance	Audit an A2A agent for spec compliance, security, and quality	protocol-engineering	Yes (or card JSON)

They compose across the software lifecycle: a2a-expose builds an exposure, a2a-conformance audits it, and a2acli operates it in production. The Worked example shows this end to end.

See also

For scenario walkthroughs, a prompt library, and authoring guidelines, the canonical source is the repository’s Agent Skills Guide (docs/SKILLS.md).

The three skills

a2acliDrive the CLI: discover, send, stream, poll, manage, authenticate.

a2a-exposeDesign-first workflow to expose a service over A2A (Go / Python).

a2a-conformanceTier 0/1/2 audit against ~110 normative rules with a report template.