Generate seven-section analysis files and a summary from parallelized research.
npx skills add https://github.com/compozy/agh --skill agent-exploration
Please help me install this Agent Skill. Skill: agent-exploration Source: https://github.com/compozy/agh/tree/main/.agents/skills/agent-exploration Command: npx skills add https://github.com/compozy/agh --skill agent-exploration
This Skill turns a broad research question into multiple non-overlapping, deeply read slices that produce written analysis artifacts instead of chat-only output.
explorer agents concurrently via compozy exec.summary.md that consolidates and compares findings across slice files.Run: compozy exec --skill agent-exploration -- --path ./research-current --agents 3 --prompt "How should we structure a local-first agent runtime to support durable, inspectable sessions?"
High-intent search queries and answers about installing and using this skill.
Parallel research dispatches multiple explorer agents concurrently to deeply read non-overlapping slices of your codebase, writing schema-validated analysis files instead of chat-only output while producing an aggregated synthesis summary.
Splitting a broad research question into independent codebase exploration tasks works best by dispatching up to 8 non-overlapping slices concurrently via agent orchestration, ensuring each explorer agent writes one schema-validated analysis file.
Yes, you can control the model and reasoning effort by forwarding runtime controls through the agent orchestration exec command, specifying IDE, model, and reasoning effort alongside the number of explorer agents dispatched.
Parallel agent exploration is designed exactly for multi-area discovery, dispatching scoped-write explorer agents that produce non-overlapping slice outputs and a parent-authored synthesis summary for later review and action.
You can dispatch up to 8 concurrent explorer agents for parallel research on a single topic, with each agent producing a non-overlapping written analysis artifact and a consolidated summary file aggregating findings.
Parallel research agents for codebase exploration are limited to 8 concurrent slices, require strict scoped-write schema compliance producing exactly one seven-section analysis per slice, and operate within the forwarded runtime controls of your selected model and reasoning effort.
evolution-foundationCoordinate parallel agent streams for oversized engineering tasks and synthesize findings into one report.
PelekeHarvest and score Reddit pain quotes using Playwright for product discovery.
rhpdsConfigure Claude Code reasoning effort levels to balance quality, latency, and token cost.
leadmind-aiCoordinate Developer phase slices with a prime_planner and three builders.