feat: sigils spec

Author: Gozala

rfc/schema-query.md

@@ -0,0 +1,82 @@
+# Schema Query Protocol: Graph Querying for Memory Protocol
+
+![draft](https://img.shields.io/badge/status-draft-yellow.svg?style=flat-square)
+
+## Editors
+
+- [Irakli Gozalishvili], [Common Tools]
+
+## Authors
+
+- [Bernhard Seefeld], [Common Tools]
+
+## Abstract
+
+This RFC extends the [Memory Protocol] with graph querying capabilities that enable complex queries across fact relationships. The schema query protocol introduces the `/memory/graph/query` capability, which provides a different querying mechanism from the basic `/memory/query` capability, including reference resolution, graph traversal, and schema-aware operations.
+
+## Language
+
+The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC 2119](https://www.rfc-editor.org/rfc/rfc2119).
+
+## Motivation
+
+The basic `/memory/query` capability in the [Memory Protocol] provides efficient fact retrieval but requires multiple roundtrips when working with references that point to other facts. Additionally, it returns complete facts even when only a subset of the data is needed.
+
+The schema query protocol reduces roundtrips by providing a single query capability that can follow cross-fact references and collect all relevant facts in a single response bundle. It uses schemas to determine which subset of fact data is actually needed, improving efficiency by excluding irrelevant fields.
+
+## Core Concepts
+
+### Reference Following
+
+Schema queries can follow cross-fact references during query execution, collecting related facts to build comprehensive result sets in a single roundtrip.
+
+### Bundled Responses
+
+Rather than requiring multiple queries to resolve cross-fact references, schema queries return bundled responses containing all relevant facts.
+
+### Schema-Based Filtering
+
+Schema queries use schemas to determine which portions of facts are actually needed. A fact may contain extensive data, but the schema specifies only the subset required for the query, reducing response size and improving performance.
+
+## Capabilities
+
+### `/memory/graph/query`
+
+*[Detailed specification to be completed]*
+
+This capability provides:
+
+- Automatic reference resolution
+- Graph traversal operations
+- Schema-aware filtering
+- Advanced aggregation functions
+
+## Integration with Cross-Fact References
+
+Schema queries are designed to work seamlessly with cross-fact references:
+
+- **References** are resolved to their target values during query execution
+- **Complex reference networks** can be traversed and queried as graph structures
+
+## Examples
+
+*[Examples to be provided in future versions]*
+
+## Implementation Requirements
+
+*[Implementation requirements to be specified]*
+
+## Future Work
+
+This specification is currently in draft status. Future versions will include:
+
+- Complete capability specification for `/memory/graph/query`
+- Detailed query language syntax
+- Performance optimization guidelines
+- Security considerations for cross-space graph queries
+- Integration examples with reference resolution
+
+[Common Tools]: https://common.tools/
+[Irakli Gozalishvili]: https://github.com/gozala
+[Memory Protocol]: ./memory.md
+