claudemem-orchestration

Multi-agent orchestration shares a single claudemem output across parallel agents through a shared session directory. Run claudemem once to index the codebase, write the structured output to a shared file, then launch agents in parallel to read from that file, eliminating redundant analysis and enabling consensus-based consolidation.

Yes. Create a session directory, generate a shared claudemem map, then launch multiple agents in parallel with role-based prompts and commands. Each agent reads the shared output and writes individual analyses, which consolidate into a final prioritized report.

Orchestration consolidates parallel agent findings using ultrathink-style analysis. After agents complete role-based investigations and write results to the session directory, merge findings into a single prioritized action list that reflects consensus across all parallel analyses.

No. Run claudemem once with the --raw flag to generate a structure map, save it to the shared session file, then all parallel agents read from that single output, avoiding redundant claudemem executions.

Orchestration establishes a unique session directory per analysis, writes structured claudemem output to shared files within it, launches agents in parallel to read shared data and write individual analyses, then consolidates results into a final report stored in the session directory.