docs: clarify agent integration boundaries

[seanbrar]

Summary

Clarifies the async boundary, application-owned transcript pattern, continuation/media replay contract, configuration ownership, and strict tool-argument behavior for agent-framework integrations. Adds a boundary test for the documented supported OpenAI transcript round trip.

Related issue

None

Test plan

• uv run pytest tests/providers/test_errors.py tests/interaction/test_continuation.py -v — 28 passed.
• uv run rumdl check docs/agent-loop.md docs/configuration.md docs/conversations-and-agents.md — passed.
• uv run mkdocs build --strict — passed.
• just check — Ruff, rumdl, mypy, and 450 non-API tests passed.

Notes

The docs keep media transport explicit: OpenAI-style text/tool transcripts can be imported, while PDFs, images, audio, video, and provider-native media remain application records reconstructed as Pollux Source values. Display reasoning is documented as output data rather than replay history.

---

• PR title follows conventional commits
• just check passes
• Tests cover the meaningful cases, not just the happy path
• Docs updated (if this changes public API or user-facing behavior)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!