Decompose large inputs and coordinate parallel sub-agents for synthesis.
npx skills add https://github.com/ruska-ai/orchestra --skill rlm-ruska-ai
Please help me install this Agent Skill. Skill: rlm Source: https://github.com/ruska-ai/orchestra/tree/main/.claude/skills/rlm Command: npx skills add https://github.com/ruska-ai/orchestra --skill rlm-ruska-ai
This Skill enables systematic handling of large, multi-file inputs by decomposing tasks, coordinating parallel sub-agents, and synthesizing results into a coherent answer. It is designed to make complex analyses feasible within finite context windows and with clear traceability across processing stages.
Use the RLM skill to analyze a large repository or document by triggering a decomposition workflow (e.g., "analyze entire repository"). The system will partition the input, spawn parallel workers, evaluate results, and synthesize a final report.
High-intent search queries and answers about installing and using this skill.
Analyzing a large codebase that exceeds a single context window requires decomposing tasks and coordinating parallel sub-agents. This hierarchical orchestration splits work into chunks, processes them in parallel, and reassembles findings into a coherent answer.
Processing long documents using parallel sub-agents involves a two-level supervisor-worker pattern. The supervisor partitions the input, spawns parallel workers to process document chunks, evaluates the results, and synthesizes a final report with chunk-level provenance and confidence scores.
Hierarchical task decomposition for multi-file project analysis works by splitting work into manageable chunks, processing them in parallel using Sonnet and Haiku models, and reassembling the findings. This approach handles inputs where single-pass analysis is insufficient.
You can analyze an entire repository by triggering a decomposition workflow that partitions the input, spawns parallel workers, evaluates results, and synthesizes a final report. This method bypasses single-pass limitations by coordinating multiple sub-agents across the codebase.
The limitations of using parallel sub-agents for large input processing include potential overhead from coordinating multiple workers and the need to synthesize cross-chunk patterns. It is designed for tasks where single-pass analysis is insufficient rather than for simple queries.
QredenceChunk large documents and codebases for Daytona RLM workspace processing.
prateekAggregate diverse sources into citation-backed research reports with credibility scoring.
hawkyrePartition pull requests into chunks and run parallel specialist audits.
Kling0012Coordinate multi-agent swarms for parallel task execution across mesh, hierarchical, and adaptive topologies.
