Babysitter research-orchestration

Parallel research agent orchestration dispatching 5-10 concurrent agents for comprehensive multi-source research with synthesis and validation.

install
source · Clone the upstream repo
git clone https://github.com/a5c-ai/babysitter
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/a5c-ai/babysitter "$T" && mkdir -p ~/.claude/skills && cp -r "$T/library/methodologies/claudekit/skills/research-orchestration" ~/.claude/skills/a5c-ai-babysitter-research-orchestration && rm -rf "$T"
manifest: library/methodologies/claudekit/skills/research-orchestration/SKILL.md
source content

Research Orchestration

Overview

Orchestrates 5-10 parallel research agents for comprehensive, multi-source research. Achieves up to 90% faster results compared to sequential research through concurrent execution.

Research Depths

DepthAgent CountUse Case
Shallow5Quick fact-finding, simple queries
Medium7Standard research, moderate complexity
Deep10Comprehensive analysis, complex queries

Process Flow

  1. Plan: Decompose query into independent sub-queries
  2. Dispatch: Run 5-10 research agents in parallel
  3. Synthesize: Merge findings, identify consensus and conflicts
  4. Validate: Cross-reference against codebase for accuracy

Sources

  • Codebase: source files, patterns, implementations
  • Documentation: README, JSDoc, inline comments
  • Configuration: package.json, tsconfig, CI/CD configs

Confidence Scoring

Overall confidence is a weighted average of individual agent confidence scores, adjusted by validation results. Below 70% triggers human review.

When to Use

  • /research [query]
    slash command
  • Before specification creation
  • When investigating unfamiliar parts of the codebase

Processes Used By

  • claudekit-research
    (primary consumer)