bootstrap-memory

Seed OpenMemory with project knowledge by indexing key prompts and active decisions while applying simhash-based per-entry deduplication checks to prevent duplicates. This ensures safe re-runs and incremental knowledge curation.

Index active decision summaries into OpenMemory by linking entries to decision IDs and storing metadata such as simhash, sector, and IDs. This maintains provenance and supports robust semantic search and traceability for future retrieval.

Simhash-based deduplication computes a fingerprint for each prompt and decision entry before storage. It checks these fingerprints against existing OpenMemory records to skip duplicates, allowing safe incremental seeding without data redundancy.

Yes, you can safely re-run knowledge curation because the seeding process applies per-entry dedup checks. It evaluates simhash fingerprints for each prompt and decision before storage, preventing duplicate records during repeated executions.

Yes, the seeding process requires reading project knowledge from key prompts located in the ~/.claude/.claude/tracking/key-prompts/ directory, along with stored active decisions, to index them into OpenMemory with appropriate metadata.

Incremental memory seeding preserves metadata tags like sector, simhash, and decision IDs to maintain data provenance. This explicit metadata tagging enables robust semantic search capabilities and ensures traceability back to original project decisions.