How adventh007/theme02-troubleshooting-engine works
The product is a REST service that turns a complaint and a supplied SIIS article into a schema-validated, guided troubleshooting response with deeplinks. The documented workflow is API intake → query cleanup and article relevance scoring → grounded article-step parsing → action structuring and deeplink matching → validation and caching. The graph follows the sampled calls and README pipeline; optional LLM refinement and semantic matching are shown as separate capabilities. Several module bodies were not sampled, so uncertain connections are omitted.
Main entry points: engine.py Workflow engine · step_structurer.py Action structurer · deeplink_matcher.py Deeplink matcher · main.py Troubleshooting API
Components 15
API
- Troubleshooting API
app/main.py
Troubleshooting flow
- Workflow engine
app/engine.py - Query enrichment
app/query_enrichment.py - Article relevance
app/relevance.py - SIIS step parser
app/siis_parser.py - Action structurer
app/step_structurer.py - Action categorizer
app/categorizer.py - Response schema
app/schema.py
Matching
- Deeplink matcher
app/deeplink_matcher.py - Deeplink catalog
app/data/deeplinks.json - Embedding model
app/embeddings.py
Optional services
- Result cache
app/cache.py - LLM refinement
app/llm_client.py
Other components
- API caller
- LLM provider
Connections 16
- API caller → Troubleshooting API: submits query and article
README.md - Troubleshooting API → Workflow engine: invokes processing
app/main.py - Workflow engine → Query enrichment: cleans query
app/engine.py - Workflow engine → Article relevance: scores article
app/engine.py - Workflow engine → SIIS step parser: parses article
app/engine.py - Workflow engine → Action structurer: builds actions
app/engine.py - Workflow engine → Response schema: validates response
app/engine.py - Workflow engine → Result cache: reads and writes
app/engine.py - Action structurer → SIIS step parser: uses parsed sections
app/step_structurer.py - Action structurer → Deeplink matcher: matches step groups
app/step_structurer.py - Action structurer → Action categorizer: categorizes actions
app/step_structurer.py - Action structurer → LLM refinement: optionally refines steps
app/step_structurer.py - Deeplink matcher → Deeplink catalog: loads entries
app/deeplink_matcher.py - Deeplink matcher → Embedding model: uses semantic scoring
app/deeplink_matcher.py - Article relevance → Embedding model: uses semantic scoring
app/relevance.py - LLM refinement → LLM provider: calls configured provider
app/llm_client.py