End of a long debugging rabbit hole. After ~10 attempts patching Graphiti's
compatibility with non-OpenAI LLMs (GLM, Qwen via SiliconFlow), every fix
uncovered another bug. Made the strategic call to abandon Graphiti entirely
for the graph-build path and write a minimal custom builder.
## Why Graphiti had to go
- Qwen 32B 32K context overflow on cumulative episode retrieval (60K-82K tokens)
- SiliconFlow Qwen 72B also 32K (not 128K as docs implied)
- GLM 4 Flash gives 128K context but triggers "20015 parameter invalid" on
Graphiti's structured output calls (logprobs in reranker, empty input in
embedder, nested dict in Neo4j writes)
- Each patch target was 1-2 levels deep in Graphiti internals
- 4 separate monkey-patches (EntityNode.save, bulk_utils, driver layer,
reranker, embedder) still couldn't cover the extract_nodes path
## New architecture
backend/app/services/custom_graph_builder.py (NEW, 348 lines):
- Uses Foresight's own LLMClient (retry + GLM→Qwen fallback + token tracker)
- 1 LLM call per chunk (Graphiti needed 4-5)
- Single prompt extracts entities + relationships as JSON
- Writes directly to Neo4j via cypher MERGE (no Graphiti dependency)
- Schema identical to what Graphiti produced — downstream get_all_nodes /
get_all_edges / zep_entity_reader all work unchanged
- ThreadPoolExecutor (10 workers) for concurrent chunk extraction
- Sequential Neo4j writes via shared session (avoids name conflicts in dedup)
- DDL wrapped in try/except (tolerates pre-existing Graphiti indexes)
- Entity name dedup via in-memory map → single uuid per canonical name
- Regex-sanitized label / relation type to prevent cypher injection
backend/app/api/graph.py:
- /api/graph/build endpoint now calls CustomGraphBuilder instead of
builder.add_text_batches() → client.add_episodes_batch() (the old Graphiti path)
backend/app/services/graph_builder.py:
- _build_graph_worker also updated to use CustomGraphBuilder (dead code path
but kept in sync)
backend/app/models/project.py:
- Default chunk_size 500 → 250 (to keep individual LLM prompts small)
backend/app/services/graphiti_client.py:
- Kept all monkey-patches for backward compat — they now only affect legacy
Graphiti code paths that CustomGraphBuilder bypasses entirely:
* _patch_neo4j_driver (AsyncSession.run nested-dict sanitize)
* _patch_reranker_for_non_openai (GLM logprobs workaround)
* _patch_embedder_empty_input (SiliconFlow empty-input guard)
* _patch_entity_node_ops (EntityNode.save sanitize)
* _patch_add_nodes_and_edges_bulk_tx (bulk-episode sanitize)
- These are kept for safety; Graphiti code path is no longer invoked in the
production build flow but methods like add_episode still exist on the class
## Performance
- Sequential: ~194 chunks × 2-5s = 10-15 min
- Concurrent (10 workers): ~194 / 10 × 3-5s = **1-2 min** (5-8x speedup)
- Rate limiting handled by LLMClient retry/backoff, not raw thread contention
## Downstream compatibility
Verified:
- get_all_nodes: MATCH (n:Entity) WHERE n.group_id = $gid RETURN n, labels(n)
- get_all_edges: MATCH (a)-[r]->(b) WHERE r.group_id = $gid RETURN r, type(r)
- get_node / get_node_edges: MATCH by uuid
CustomGraphBuilder writes:
- (n:Entity [optional second label]) with uuid, name, summary, group_id, created_at
- [r:RELATION_TYPE] with uuid, name, fact, group_id, created_at
Schema matches exactly.
## Pipeline wiring
Step 1 ontology generation → Step 2 graph build (CustomGraphBuilder) →
Step 3 profile generation (ZepEntityReader queries Neo4j) → Step 4 simulation
No changes needed downstream of Step 2.
- Merge 41 upstream commits (i18n for 7 languages, security fixes, new features)
- Rebrand all MiroFish references to Foresight/先见之明 across 37 files
- Re-apply dark theme CSS overhaul (pure black/gray, no blue tints)
- Re-apply Teleport-based theme toggle (inline with brand, 20px gap)
- Restore Foresight logo and favicon
- Update GitHub links to liyizhouAI/foresight
- Update locale files, README, package.json
Generated with [Claude Code](https://claude.ai/code)
via [Happy](https://happy.engineering)
Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Happy <yesreply@happy.engineering>
Background threads (graph building, simulation prep, report generation,
profile generation) now inherit the requesting user's locale preference.
Previously these fell back to 'zh' because Flask request context was
unavailable in spawned threads.
- Updated the `get_graph_data` method in `graph_builder.py` to include additional attributes such as creation time, validity periods, and episodes for nodes and edges, improving the richness of the graph data.
- Modified the detail panel in `Process.vue` to present new attributes, including properties, episodes, and timestamps, enhancing user interaction and data visibility.
- Improved styling for the detail panel to better organize and present the comprehensive information for selected nodes and edges.
- Updated `run.py` to conditionally print startup information only in the reloader process to avoid duplicate logs in debug mode.
- Modified `__init__.py` to log startup and completion messages based on the reloader process condition.
- Added warnings suppression in `graph_builder.py` for Pydantic v2 regarding Field usage.
- Revised `ontology_generator.py` to enforce strict design guidelines for entity types and relationships, ensuring compliance with new requirements.
- Improved logging behavior in `logger.py` to prevent log propagation to the root logger, avoiding duplicate outputs.
liyizhouAI
ghostubborn
666ghj