Your agent remembers decisions made weeks ago
Missions, intent logs, and ADRs live in SQLite. Every session starts with a full briefing — no re-explaining context, no lost reasoning.
Every new chat starts from zero. Your agent has no memory of what was decided last week.
Past ~200 files, keyword search misses cross-file impact. You need a real dependency graph.
Switch IDEs, switch sessions, onboard a new agent — the accumulated understanding evaporates.
npm install -g @precisionutilityguild/liquid-shadowOne install registers the MCP server system-wide. No per-project packages.
liquid-shadow initRegisters the MCP server and injects skills into your agent environment.
/onboardRun /onboard in any repo. Liquid Shadow indexes it, builds the graph, and activates tactical awareness.
Missions, intent logs, and ADRs live in SQLite. Every session starts with a full briefing — no re-explaining context, no lost reasoning.
When files move or code changes, NanoRepair heals the links. The "why this code exists" survives any rename or restructure.
One call returns circular dependencies, dead code, and full blast radius — not possible with grep. Real graph traversal, not pattern matching.
Fuse separate repos into a single workspace for cross-repo impact analysis and API contract detection. No copying, no manual linking.
Signatures, summaries, and structure — built for narrow context windows. Your agent gets exactly what it needs, nothing it doesn't.
Cross-language flow tracing in a single unified index. Call chains that cross language boundaries traced end-to-end.