Metadata-Version: 2.4
Name: algovoi-keystone-qdrant
Version: 0.1.0
Summary: Bind Qdrant vector writes to the AlgoVoi keystone -- wrap any QdrantClient and every upsert / delete / payload write gets a content-addressed execution_ref (keystone-only edition, Apache-2.0)
Author-email: AlgoVoi <chopmob@gmail.com>
License: Apache-2.0
Keywords: qdrant,vector-database,embeddings,agent-memory,keystone,execution_ref,jcs,rfc8785,algovoi
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Database
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: NOTICE
Requires-Dist: algovoi-execution-ref>=0.1.0
Dynamic: license-file

# algovoi-keystone-qdrant

Bind Qdrant vector writes to the AlgoVoi keystone. Agent memory is a data plane: wrap any `QdrantClient` and every `upsert` / `delete` / `set_payload` / `overwrite_payload` is bound to the `decision_ref` that authorised it, emitting a content-addressed `execution_ref` keyed to the collection. Reads (search, retrieve, scroll, query) pass through unbound; no qdrant-client dependency.

```
execution_ref = "sha256:" + SHA-256(JCS({ decision_ref, action_type, scope, outcome, executed_at_ms }))
```

Records emitted on `.executions` drop straight into `algovoi-keystone-validate`.

Keystone-only edition, Apache-2.0. Built on the AlgoVoi substrate (RFC 8785 JCS + SHA-256).
