ThothII/.superpowers/sdd/2026-08-08-internal-qdrant-ollama/task-5-report.md

Task 5 Report — Implement the Qdrant VectorStore adapter

Status

Implemented on 2026-08-08 in /Users/mp/projects/ThothII/.worktrees/git-workspace-registry.

RED evidence

Command:

cd harness
./.venv/bin/pytest tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py -q

Observed before implementation:

• exit code 2
• collection failed during import because the adapter did not exist yet

Representative failures:

• ModuleNotFoundError: No module named 'tht.adapters.vector.qdrant'

GREEN evidence

Focused behavior suite:

cd harness
./.venv/bin/pytest tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py -q

• exit code 0
• 31 passed, 1 warning

Touched-file lint:

cd harness
./.venv/bin/ruff check tht/adapters/vector/qdrant.py tht/adapters/vector/__init__.py \
  tht/ports/vector.py tht/vectorstore/records.py tht/vectorstore/store.py \
  tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py

• exit code 0
• All checks passed!

Patch hygiene:

git diff --check

• exit code 0

What changed

• added QdrantVectorStore with direct requests-based REST calls for:
  • GET /collections/{collection}
  • PUT /collections/{collection}
  • PUT /collections/{collection}/index
  • PUT /collections/{collection}/points?wait=true
  • POST /collections/{collection}/points/query
  • POST /collections/{collection}/points/scroll
  • POST /collections/{collection}/points/delete?wait=true
• implemented idempotent collection provisioning for 1024 dimensions and Cosine distance
• created deterministic UUIDv5 point IDs from workspace, semantic kind, and canonical record key
• preserved canonical record identity and only upserted/deleted points matching the exact workspace and generation filters
• added Qdrant payload helpers so stored payloads carry:
  • workspace_id
  • grouped semantic kind (schema, evidence, memory)
  • original record_kind
  • canonical record_key
  • content_hash
  • existing Thoth metadata fields
• mapped Qdrant payloads back into existing VectorHit objects without losing the original Thoth kind
• exported the new adapter from the public vector adapter package and added focused contract tests
• sanitized timeout and malformed-response failures so CLI-facing callers do not leak raw endpoint details

Self-review

• confirmed collection mismatch fails without any delete/recreate path
• confirmed every query/scroll/delete operation includes a workspace filter
• confirmed the adapter never deletes or rewrites unrelated Qdrant points
• added keyword payload indexes for all filter-critical fields used here, including document_id for exact Evidence filtering

Concerns

• the requested adversarial-review skill could not run its full external reviewer flow in this environment because the skill’s referenced brain/ files are missing at /Users/mp/.agents/skills/adversarial-review; I performed a manual adversarial self-review instead
• the focused suite still emits one pre-existing warning from testcontainers.postgres

Fix round 1 — 2026-08-08

Findings addressed

• IMPORTANT: metadata collisions could override canonical Qdrant payload identity fields and break workspace isolation
• IMPORTANT: scroll-based operations only read the first page and did not follow next_page_offset, making existing_hashes, list_evidence_generations, and delete counts inexact beyond one page

RED evidence

Command:

cd harness
./.venv/bin/pytest tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py -q

Observed before the fix:

• exit code 1
• 2 failed, 31 passed, 1 warning

Representative failures:

• assert payload["workspace_id"] == "demo" failed because colliding record.metadata overwrote canonical payload fields
• paginated scroll test missed later pages, so existing_hashes and generation cleanup counts were incomplete

GREEN evidence

Command:

cd harness
./.venv/bin/pytest tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py -q

Observed after the fix:

• exit code 0
• 33 passed, 1 warning

Touched-file lint:

cd harness
./.venv/bin/ruff check tht/adapters/vector/qdrant.py tht/vectorstore/records.py \
  tests/test_qdrant_vector_store.py tests/test_vector_port_contract.py

• exit code 0
• All checks passed!

Patch hygiene:

git diff --check

• exit code 0

Minimal fix

• made qdrant_payload apply canonical fields after record.metadata so workspace ID, semantic kind, original record kind, canonical record key, and content hash cannot be overridden by metadata collisions
• paginated _scroll until next_page_offset is absent, sent the returned offset back on the next request, and reject repeated offsets as malformed to avoid infinite loops