Home > Tasks > Reasoning > StructuralInvariantAnalysisTask

StructuralInvariantAnalysisTask

Distill complex objects down to their immutable properties and symmetries through rigorous decontextualization and mathematical stress testing. Proves structural integrity across domain boundaries.

Category: Reasoning Model: GPT-4 Preferred Side-Effect: Safe

⚙️ TaskConfig.json


{
  "task_type": "StructuralInvariantAnalysisTask",
  "subject_object": "Distributed Consensus",
  "transformation_types": [
    "network_partition",
    "latency_scaling"
  ],
  "output_format": "signature",
  "input_files": ["src/consensus/Raft.kt"]
}

→

👁️ User UI (Result Tab)

Overview

Prompt

Analysis

Result

Decontextualized Description

A mechanism for achieving agreement on a single data value among distributed processes despite failures...

Identified Invariants

Quorum Consistency: Property persists under network partition.
Monotonicity: State progression remains constant across scaling.

Live Results Showcase

Explore actual artifacts generated by this task. The file tree below points to the StructuralInvariantAnalysisTask workspace, showing transcripts and analysis results from recent test runs.

Execution Configuration

Field	Type	Description
`subject_object` *	String	The object to analyze (e.g., 'A Prime Number', 'Distributed Consensus').
`transformation_types`	List<String>	Transformations to apply. Defaults: `symmetry_groups`, `limit_cases`, `context_inversion`.
`output_format`	String	`fingerprint` (list of invariants) or `signature` (hashable summary).
`input_files`	List<String>	Specific files or patterns to be used as primary input context.
`related_files`	List<String>	Additional files for secondary context.

Token Usage: Medium (Context dependent)

Task Lifecycle

Initialization: Validates subject_object and output_format.
Context Loading: Aggregates content from input_files and priorCode.
Decontextualization: LLM strips domain terminology to find the mathematical core.
Stress Testing: Applies transformations (scaling, rotation, permutation) to test property stability.
Invariant Extraction: Identifies properties that survive all transformations.
Finalization: Generates a transcript and renders the result in the Session UI.

Error Handling

The task performs pre-execution validation on configuration fields. If the LLM fails to identify invariants, the task logs the failure in the transcript and provides a partial analysis of the stress test results.

Orchestration Boilerplate

To invoke this task headlessly using the UnifiedHarness (as detailed in the Embedding Guide):


import com.simiacryptus.cognotik.plan.tools.reasoning.StructuralInvariantAnalysisTask
import com.simiacryptus.cognotik.plan.tools.reasoning.StructuralInvariantAnalysisTask.StructuralInvariantAnalysisTaskExecutionConfigData
import com.simiacryptus.cognotik.plan.tools.TaskTypeConfig

fun runAnalysis(harness: UnifiedHarness, projectDir: File) {
    // 1. Define Runtime Input
    val executionConfig = StructuralInvariantAnalysisTaskExecutionConfigData(
        subject_object = "Distributed Consensus",
        transformation_types = listOf(
            "network_partition", 
            "latency_scaling", 
            "context_inversion"
        ),
        output_format = "signature",
        input_files = listOf("src/consensus/*.kt")
    )

    // 2. Execute via Harness
    harness.runTask(
        taskType = StructuralInvariantAnalysisTask.StructuralInvariantAnalysis,
        typeConfig = TaskTypeConfig(), // Use default static settings
        executionConfig = executionConfig,
        workspace = projectDir,
        autoFix = true
    )
}

Prompt Segment

The following logic is injected into the orchestrator's context:


StructuralInvariantAnalysis - Distill an object to immutable properties
  ** Specify the subject_object to analyze
  ** Define transformation_types (e.g., symmetry_groups, limit_cases)
  ** Select output_format ('fingerprint' or 'signature')
  ** Process involves:
     - Decontextualization (stripping domain terminology)
     - Stress Testing (applying transformations)
     - Invariant Extraction (identifying constants)
     - Signature Generation