Skip to content
a2acli

Skill: a2a-conformance

Category: protocol-engineering · Source: skills/a2a-conformance/SKILL.md

What it does

Section titled “What it does”

The a2a-conformance skill evaluates any A2A Protocol v1.0 remote agent for normative spec compliance, multi-transport behavioral equivalence, security, and quality design, then generates a structured audit report with stable check IDs. It activates on prompts about auditing an A2A agent, running conformance checks, evaluating a service against the A2A spec, or checking compatibility.

How an agent uses it

Section titled “How an agent uses it”

The skill applies a multi-tier assessment, escalating with the access level available:

  • Tier 0 — Static. Inspect the Agent Card JSON and SDK source; no server required. Validates required card fields against the A2A-CARD-* rules.
  • Tier 1 — Probes. Run the automated a2acli black-box probe battery against a live URL and record PASS / WARN / FAIL per A2A-* ID.
  • Tier 2 — TCK. Orchestrate the official a2aproject/a2a-tck pytest suite via Python/uv for a formal compatibility result.

It follows four core principles: RFC 2119 strictness (any MUST failure is blocking), the tiered escalation above, stable check IDs, and actionable, copy-pasteable remediation. Verdicts are:

  • CONFORMANT — 100% of MUST checks pass, 0 MUST failures.
  • CONFORMANT WITH WARNINGS — 100% MUST pass, but SHOULD warnings exist.
  • NOT CONFORMANT — one or more MUST checks fail.

A typical Tier 1 run:

Terminal window
# Automated probe battery
./skills/a2a-conformance/scripts/probe.sh http://127.0.0.1:9001


# Key manual probes
a2acli discover --service-url http://127.0.0.1:9001 --output json
a2acli conformance --service-url http://127.0.0.1:9001 --output json
a2acli list tasks --service-url http://127.0.0.1:9001 --limit 5 --output json

What it ships

Section titled “What it ships”
  • references/ — checklist (the master normative checklist, ~110 rules across 9 areas with stable A2A-<AREA>-<NNN> IDs and spec citations), static-review (Tier 0), probes (Tier 1 a2acli battery), tck (Tier 2), and scoring (verdict rules and worked example).
  • assets/report.template.md — the conformance report template.
  • scripts/ — probe.sh (Tier 1) and run-tck.sh (Tier 2).

Check IDs are grouped by area: A2A-CARD-*, A2A-OPS-*, A2A-TASK-*, A2A-STREAM-*, A2A-PUSH-*, A2A-BIND-*, A2A-VER-*, A2A-SEC-*, A2A-EXT-*, and A2A-QUAL-* (non-normative quality).