Use real Qdrant backend in tests

[Teagan42]

What

• document version bump process in AGENTS.md
• run tests against a real Qdrant instance using actual embedding models
• fix server Qdrant queries for scroll, search and recommend APIs
• add sentence-transformers to dev dependencies

Why

• ensure contributors consistently bump project versions
• exercise the full Qdrant stack in tests instead of dummies
• align server code with current qdrant-client API
• provide reranker dependency for test environment

Affects

• developer docs
• test suite
• server Qdrant integration

Testing

• uv run ruff check .
• uv run pytest

Documentation

• Updated AGENTS.md

---

https://chatgpt.com/codex/tasks/task_e_68bf9d3be7808328bc267f692ea47659

[github-actions]

Coverage Report

File	Stmts	Miss	Cover	Missing
mcp_plex
loader.py	186	16	91%	32–34, 301, 304–307, 352–361, 599, 605
server.py	190	22	88%	75, 93–94, 205–221, 313, 317, 337–341, 374
TOTAL	478	38	92%

Tests	Skipped	Failures	Errors	Time
19	0 💤	0 ❌	0 🔥	16.993s ⏱️