Issue #4632: Invocation ID collision-resistance contract coverage

[davidahmann]

Problem observed

Invocation IDs are the primary correlation key for ADK run lifecycle events, yet there was no direct regression test asserting UUID-backed uniqueness at generation boundaries. Without an explicit contract test, future refactors could unintentionally reduce entropy or alter format assumptions and weaken event/session traceability under concurrent invocations.

Why it matters operationally

Operators and downstream tooling rely on invocation IDs to correlate events, tool callbacks, and replay context. If identifier quality degrades, debugging and audit workflows become noisy and ambiguous. Capturing this behavior in a focused unit test keeps the contract visible and protects reliability without introducing runtime behavior changes.

Minimal repro

# Generate invocation IDs repeatedly from the helper.
uv run python - <<PY
from google.adk.agents.invocation_context import new_invocation_context_id
print([new_invocation_context_id() for _ in range(10)])
PY

Fix approach

A new unit test validates three critical properties of generated invocation IDs: required e- prefix, UUID parseability for the suffix, and uniqueness across repeated generations. This test intentionally focuses on contract semantics rather than implementation details, keeping room for internal refactors while guarding collision-resistance guarantees.

Validation evidence

• uv run pytest tests/unittests/agents/test_invocation_context.py -k "new_invocation_context_id_is_uuid_backed_and_unique" passed.
• The assertion verifies uniqueness across multiple generated values.
• The assertion verifies every value conforms to expected prefix + UUID structure.

Open follow-up question for maintainers

Would maintainers want a companion test at the runner layer that validates generated invocation IDs remain unique across equivalent normalized prompt inputs in end-to-end invocation setup?

This contribution was informed by patterns from Wrkr. Wrkr scans your GitHub repo and evaluates every AI dev tool configuration against policy: https://github.com/Clyra-AI/wrkr