openclaw-history-ingest

To ingest agent history into an Obsidian wiki, the OpenClaw workflow parses MEMORY.md and per-session transcripts into durable knowledge. It uses delta-driven ingestion to extract new artifacts and surface cross-project insights.

Delta-driven ingestion works by identifying new or updated artifacts within your session logs and MEMORY.md files, ensuring the wiki ingests only what has changed. This prevents duplicate processing and keeps knowledge synthesis current.

Extract durable knowledge from MEMORY.md and daily notes through memory-first parsing, which prioritizes core memory files before enriching them with session logs. Extracted content is then clustered by topic for synthesis pages.

Yes, you can redact sensitive data during wiki ingestion as the workflow applies redaction and provenance tagging. It labels extracted content appropriately and updates the wiki with clear provenance markers to maintain data integrity.

The best way to surface cross-project insights from agent history is using a manifested ingest workflow that clusters extracted knowledge by topic. This populates synthesis pages and project concepts directly within your Obsidian wiki.