john/ember-memory-test
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat(episode): BACKFILL_q-10826044-0af5-4aa9-ae8e-6e4797032ec4_S20260510.BACKFILL-RECOVERED_XX.bkf.0.warm_qdr.orphan.in.cc.es.000.BKF_J.01BBB_E.CAB-3

Browse source 
Skill: NONE | Type: chronicle
Summary: This episode deliberately mentions 12 canonical entities from the EMBER ecosyste
This commit is contained in:
Ember 2026-05-11 04:33:41 +00:00
parent 064d6dac57
commit 593133ffb9
1 changed files with 27 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
episodes/BACKFILL_q-10826044-0af5-4aa9-ae8e-6e4797032ec4_S20260510.BACKFILL-RECOVERED_XX.bkf.0.warm_qdr.orphan.in.cc.es.000.BKF_J.01BBB_E.CAB-3.md Normal file
View file

@ -0,0 +1,27 @@
---
episode_id: "01bbbcab-3e09-5aea-b2e9-94aa6c1e891f"
puente_flat: "BACKFILL_q-10826044-0af5-4aa9-ae8e-6e4797032ec4_S20260510.BACKFILL-RECOVERED_XX.bkf.0.warm_qdr.orphan.in.cc.es.000.BKF_J.01BBB_E.CAB-3"
session_code: "S20260510.BACKFILL_RECOVERED"
acto_dialogico: "informar"
actor_flujo: "ember_internal"
criticidad_aegis: "moderate"
skill_dominante: "NONE"
fase_proyecto: "maintenance"
tipo_semantico: "chronicle"
summary_one_line: "This episode deliberately mentions 12 canonical entities from the EMBER ecosystem to stress-test the Neo4j MENTIONS edge creation in CRISOL F5b. Each entity listed in entidades_canonicas should produc"
source_type: "claude_code"
trust_boundary: "default"
created_at: "2026-05-10T00:00:00+00:00"
relectura_tagged: false
forgejo_commit_sha: "pending"
---
This episode deliberately mentions 12 canonical entities from the EMBER ecosystem to stress-test the Neo4j MENTIONS edge creation in CRISOL F5b. Each entity listed in entidades_canonicas should produce one MENTIONS edge connecting the episode node to the entity node in the Neo4j graph.
The 12 entities and their roles in the CRISOL pipeline are as follows. PostgreSQL serves as the primary relational store (F5 target) where episode metadata is persisted in the episodes table. Qdrant provides the vector store (F5a target) where episode embeddings enable semantic search. Neo4j hosts the knowledge graph (F5b target) where entity relationships are materialized as typed edges. Nextcloud functions as the document store (F5c target) where episode markdown files are backed up for human access.
Forgejo provides the Git-based version control for CRISOL pipeline code, enabling change tracking and rollback. LiteLLM acts as the unified LLM gateway that routes embedding requests to Gemini 001 for vector generation. The servidor MEMORIA CX53 hosts all these services as Docker containers on the Hetzner infrastructure. PUENTE is the molecular identifier system that generates unique episode IDs encoding 21 dimensions.
TESSERA defines the episode format specification including EJES_EPISODICOS, TRUST_BOUNDARIES, and PREGUNTAS_HyDE sections. CRISOL is the ingestion pipeline itself (F0 through F5) that processes episodes from inbox to all four stores. LEXICON is the ontology that maps entity names to canonical identifiers with aliases, definitions, and relationships. The episode_classifier.py module in CRISOL performs mode classification (R1-R11 rules) to determine the semantic type of each episode.
When F5b processes this episode, it should create 12 MENTIONS edges in Neo4j. For each entity, F5b performs a LEXICON lookup to verify the entity exists as an active entry, then executes a MERGE to create or match the entity node, then creates the MENTIONS edge with properties including confidence (1.0 for entities explicitly listed in entidades_canonicas) and episode_id.