claudemem-orchestration

What problem does it solve?

This Skill coordinates multi-agent code analyses by sharing claudemem outputs across parallel agents, enabling parallel investigations and consensus-based prioritization while avoiding redundant claudemem runs.

Core Features & Use Cases

• Single claudemem run: Run claudemem ONCE to get structural overview and index the codebase, then reuse the results across agents.
• Shared output: Write output to a shared file in the session directory so all agents read from the same source.
• Parallel investigations: Launch multiple agents in parallel, each reading the shared claudemem output.
• Consensus consolidation: Consolidate results with ultrathink-style analysis into a final report.

Quick Start

1. Create a unique session directory for the orchestration.
2. Run claudemem --nologo map "query area" --raw to generate a structure map and save it to the shared file.
3. Launch agents in parallel that read the shared file and write individual analyses.
4. Consolidate findings into a final consolidated report in the session directory.