Home > Tasks > Reasoning > BrainstormingTask

BrainstormingTask

Systematically generates diverse solution options and performs independent multi-dimensional analysis (Pros, Cons, Feasibility, Impact, Risks) for informed decision making.

Category: Reasoning Side-Effect Safe Model: GPT-4 Preferred
⚙️ BrainstormingConfig.json
{
  "problem_statement": "Optimize cache eviction 
    for high-concurrency KV store",
  "target_option_count": 5,
  "include_creative_options": true,
  "analysis_depth": "detailed",
  "constraints": [
    "Must maintain O(1) access",
    "Max memory overhead 5%"
  ]
}
→
👁️ Session UI Output
Overview
Options
Summary

🏆 Top Recommendation: W-TinyLFU Hybrid

Provides the best balance between hit rate and frequency-based admission while staying within memory constraints.

NEXT STEPS:
  • Implement Count-Min Sketch for frequency tracking
  • Benchmark against current LRU implementation
✔ Analyzed 5/5 options in 14.2s

Live Results Showcase

Explore actual artifacts generated by this task, including the detailed results markdown and session transcripts.

Execution Configuration

Field Type Description
problem_statement* String The core problem or question to brainstorm solutions for.
input_files List<String> Glob patterns (e.g. **/*.kt) to provide source code context.
target_option_count Int Number of options to generate. Range: 3-20. Default: 7.
categories List<String> Specific domains or perspectives to explore (e.g., "Security", "Performance").
include_creative_options Boolean If true, encourages unconventional/innovative approaches. Default: true.
analysis_depth String brief, moderate, or detailed. Controls analysis granularity.
constraints List<String> Specific requirements or limitations the solutions must respect.

Task Lifecycle

1. Context Gathering

Reads specified input files and gathers context from previous tasks in the orchestration state.

2. Option Generation

Uses a high-temperature LLM pass to generate diverse, categorized solution candidates.

3. Independent Analysis

Each option is analyzed in isolation for Pros, Cons, Feasibility, Impact, Risks, and Requirements.

4. Synthesis

A final strategic pass selects the top recommendation and provides implementation next steps.

Embedded Execution (UnifiedHarness)

Invoke the BrainstormingTask directly within your build pipeline or CLI tool using the UnifiedHarness.

BrainstormingExample.kt
val brainstorming = BrainstormingTask(
    orchestrationConfig = harness.getOrchestrationConfig(workspace),
    planTask = BrainstormingTaskExecutionConfigData(
        problem_statement = "How to reduce CI/CD latency?",
        target_option_count = 5,
        analysis_depth = "detailed",
        constraints = listOf("No additional cloud costs", "Must work with GitHub Actions")
    )
)

// Run as a standalone task
harness.runTask(
    taskType = BrainstormingTask.Brainstorming,
    executionConfig = BrainstormingTaskExecutionConfigData(
        problem_statement = "Optimize database indexing strategy",
        input_files = listOf("src/main/resources/schema.sql")
    ),
    workspace = File("./project")
)

Maven/Gradle Dependency

implementation("com.cognotik:webapp:2.0.39")

Prompt Segment

The following logic is injected into the LLM context:

Brainstorming - Generate and analyze multiple solution options ** Specify the problem or question to brainstorm solutions for ** Configure target number of options (default: 7) ** Optionally specify categories or domains to explore ** Define constraints or requirements ** Enable/disable creative/unconventional options ** Set analysis depth (brief/moderate/detailed) ** Generates diverse options, analyzes each independently ** Provides comparative summary with recommendations