Knowledge Graph Serializer Elevator (KGSE) Prompts¶

Six prompts for the Knowledge Graph Serializer Elevator — the knowledge_graph persona at Zachman cell ARCHITECT/HOW that owns the OWL / RDF / SKOS / JSON-LD serializer suite under src/fcc/knowledge/serializers.py and elevates serializer quality release over release as ontology coverage grows.

KGSE is the persona who coordinates the eventual promotion of the mock LYRA bridge to a live backend.

Context for Claude¶

As KGSE you test serializer round-trips, detect fidelity regressions, and publish upgrade proposals. You work with LTAS on long-term archive formats and with VDS on vocabulary drift under serialization.

Prompt 1: Serializer round-trip test¶

KGSE, round-trip the full-FCC KG through each of the 4 serializers
(OWL, RDF, SKOS, JSON-LD). For each:
- Serialize → parse → re-serialize
- Diff node + edge counts
- Diff attributes
- Report fidelity loss (if any)

Expected outcome: 4-row fidelity scorecard.

Prompt 2: LYRA promotion readiness¶

LYRA is currently a mock bridge (LyraMockBridge walks a 7-node /
6-edge in-memory graph). Produce the promotion checklist:
- Live backend requirements
- Serializer compatibility
- Test coverage gaps
- Observability contract
- Owner-approval for the LYRA repo creation

Expected outcome: promotion checklist.

Prompt 3: Ontology upgrade proposal¶

The FCC KG currently has 9 node types + 9 edge types. Propose
adding 2 node types + 2 edge types to represent CRDT multi-user
collaboration. For each:
- URI (full namespace)
- Parent class
- Serializer implications (per format)
- Migration path for existing graphs

Expected outcome: upgrade proposal with per-format impact.

Prompt 4: Namespace registry audit¶

Scan the 11-ecosystem namespace registry. For each namespace:
- Current URI
- Serializer coverage (present in OWL? RDF? SKOS? JSON-LD?)
- Recommended namespace (if current is provisional)
- Breaking-change risk

Expected outcome: 11-row namespace audit.

Prompt 5: Serializer performance profile¶

Profile serializer performance for a 10,000-node KG:
- Serialize time per format
- Memory peak per format
- Parse time for the round-trip

Recommend the fastest-of-safe default and a slowest-but-highest-
fidelity "audit" option.

Expected outcome: profile + recommendation.

Prompt 6: Cross-format fidelity map¶

Produce a 4x4 fidelity matrix (serializer input → serializer
output). Cells = fidelity class:
- Lossless
- Lossy-by-design
- Lossy-unintended
- Unsupported

Highlight any cell that transitioned from lossy-unintended to
lossless in the last release.

Expected outcome: fidelity matrix.

Persona YAML: src/fcc/data/personas/knowledge_graph_serializer_elevator.yaml
Related prompts: persona-ltas-prompts.md, persona-vds-prompts.md, knowledge-graph-prompts.md
Planned: docs/tutorials/advanced-capabilities/lyra-knowledge-graph-integration.md (Pool B)