How rickingithub/pkm_pub works
pkm_pub is a script-driven personal knowledge publishing system: it captures material into a vault, processes raw notes into knowledge and methodology, produces and reviews outputs, and supports publishing discovery and operational checks. The main documented path is capture → ETL/refinement → abstraction and production → human-approved feedback and audit. Separate Gemini distillation and JD-capture workflows are also present. The README gives no architectural detail, and several stage implementations were not sampled, so uncertain internal links are omitted.
Main entry points: run_full_oil_pipeline.py Stage orchestrator · etl_watcher.py ETL coordinator · search_publishing.py Publishing search · pipeline_status.py Pipeline status
Components 23
Capture
- Webhook intake
scripts/ops/webhook_server.py - JD capture
scripts/jobs/jd_fetch.py - Raw vault
Knowledge processing
- ETL entry
scripts/pipeline/run_etl_pipeline.py - ETL coordinator
scripts/pipeline/etl_watcher.py - Classify records
scripts/pipeline/classify_records.py - Vault migration
scripts/pipeline/migrate_to_arch.py - MOC network
scripts/moc/build_moc_network.py - Source scan
scripts/pipeline/source_scan.py - Stage orchestrator
scripts/pipeline/run_full_oil_pipeline.py - Refine and ETL
scripts/pipeline - Methodology abstraction
scripts/pipeline - Feedback loop
scripts/pipeline/feedback_loop.py - Gemini distillation
scripts/gemini/gemini_pipeline.py - Knowledge vault
Publishing
- Output production
scripts/pipeline - Published outputs
- Publishing search
scripts/ops/search_publishing.py
Operations
- Pipeline status
scripts/pipeline/pipeline_status.py - Audit tools
scripts/audit - Service control
scripts/ops/toggle_services.py
Other components
- Vault user
- Messaging bots
Connections 30
- Messaging bots → Webhook intake: send messages
- Webhook intake → Raw vault: write inbox
- Vault user → JD capture: submit URL
- JD capture → Raw vault: archive capture
- Vault user → ETL entry: run ETL
- ETL entry → ETL coordinator: invoke run_etl
- ETL coordinator → Classify records: classify records
- ETL coordinator → Vault migration: migrate vault
- ETL coordinator → MOC network: build network
- ETL coordinator → Source scan: scan sources
- Raw vault → ETL coordinator: read pending notes
- Vault user → Stage orchestrator: select stages
- Stage orchestrator → Refine and ETL: dispatch stage
- Stage orchestrator → Methodology abstraction: dispatch stage
- Stage orchestrator → Output production: dispatch stage
- Stage orchestrator → Feedback loop: dispatch stage
- Stage orchestrator → Audit tools: dispatch stage
- Refine and ETL → Knowledge vault: write knowledge
- Methodology abstraction → Knowledge vault: derive methodology
- Output production → Published outputs: materialize outputs
- Feedback loop → Knowledge vault: approved return
- Vault user → Gemini distillation: run distillation
- Gemini distillation → Knowledge vault: distill and polish
- Vault user → Publishing search: search publishing
- Publishing search → Published outputs: read final drafts
- Publishing search → Knowledge vault: read support cards
- Vault user → Pipeline status: request status
- Pipeline status → Raw vault: count pending
- Pipeline status → Knowledge vault: inspect inventory
- Vault user → Service control: toggle services