Skip to content

armgit / ruflo

Public Imported from GitHub
0

🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

Checkpoint: File edits

rUvf0104a8ebc5b
diff --git a/v3/implementation/adrs/ADR-051-infinite-context-compaction-bridge.md b/v3/implementation/adrs/ADR-051-infinite-context-compaction-bridge.md
index 40cc2e43de..621d7dcfa0 100644
--- a/v3/implementation/adrs/ADR-051-infinite-context-compaction-bridge.md
+++ b/v3/implementation/adrs/ADR-051-infinite-context-compaction-bridge.md
@@ -759,16 +759,16 @@ All capabilities confirmed working (2026-02-10):
## Future Enhancements
-1. **Compaction summary capture**: After compaction, store Claude's own summary as a
- high-confidence semantic entry alongside our chunk-level detail
+1. **Smarter text pruning**: Use extractive summarization instead of simple truncation
+ for old text blocks — preserve key decisions and reasoning
2. **Cross-session search MCP tool**: Expose `transcript-archive` search as an MCP
tool so Claude can explicitly query past conversations
3. **MemoryGraph integration**: Add reference edges between sequential chunks for
PageRank-aware retrieval (ADR-049)
-4. **Real ONNX embeddings**: Replace hash embeddings with ONNX-based embeddings for
- true semantic similarity search (transparent upgrade)
-5. **Adaptive retention**: Dynamically adjust retention period based on storage usage
- and access patterns
+4. **Adaptive pruning thresholds**: Dynamically adjust `TEXT_PRUNE_KEEP` and
+ `TEXT_PRUNE_THRESHOLD` based on conversation complexity and context growth rate
+5. **Upstream contribution**: Propose text pruning as a native Claude Code feature
+ to eliminate the need for SDK patching
## Implementation Details