Package neqsim.util.agentic

package neqsim.util.agentic

Agentic engineering infrastructure for NeqSim.

Provides session tracking, feedback collection, and result validation for AI agent workflows that solve engineering tasks using NeqSim's thermodynamic and process simulation API.

Key Classes:

• AgentSession — Tracks a single agent workflow session from scope through reporting, recording phases, tool invocations, and simulation runs
• AgentFeedbackCollector — Aggregates metrics across sessions for continuous improvement: success rates, failure patterns, API gaps
• TaskResultValidator — Validates task results.json files against the expected schema for the task-solving workflow

Architecture:

Agent Session Lifecycle:

  SCOPE --> RESEARCH --> ANALYSIS --> VALIDATION --> REPORTING
    |          |            |             |              |
    +----------+------------+-------------+--------------+
                        |
                  AgentSession (tracks phases, tools, simulations)
                        |
                  AgentFeedbackCollector (aggregates across sessions)
                        |
                  TaskResultValidator (validates output schema)

Version:

1.0

Author:

Even Solbraa

Related Packages

Package	Description
neqsim.util

Class	Description
AgentBenchmarkSuite	Defines and evaluates standardized engineering benchmark problems for agent performance measurement.
AgentBenchmarkSuite.BenchmarkProblem	A single benchmark problem with expected reference solution.
AgentBenchmarkSuite.BenchmarkReport	Aggregate report for the full benchmark suite evaluation.
AgentBenchmarkSuite.Difficulty	Difficulty level of the benchmark problem.
AgentBenchmarkSuite.ProblemCategory	Category of engineering benchmark problem.
AgentBenchmarkSuite.ProblemResult	Result of evaluating a single benchmark problem.
AgentFeedbackCollector	Collects and aggregates feedback metrics from agent sessions for continuous improvement.
AgentFeedbackCollector.APIGap	Record of a discovered API gap.
AgentFeedbackCollector.FailureCategory	Failure category for classification of session failures.
AgentFeedbackCollector.SessionSummary	Summary of a recorded session (lightweight copy for storage).
AgentSession	Tracks an AI agent workflow session from start to completion.
AgentSession.Outcome	Final outcome of the agent session.
AgentSession.Phase	Workflow phase in the task-solving lifecycle.
AgentSession.PhaseRecord	Record of a single workflow phase with timing.
AgentSession.SimulationRun	Record of a simulation run.
AgentSession.ToolInvocation	Record of a tool invocation.
SimulationQualityGate	Automated quality gate for simulation results validation.
SimulationQualityGate.QualityIssue	Represents a single quality gate issue.
SimulationQualityGate.Severity	Issue severity levels.
TaskResultValidator	Validates task results.json files against the expected schema for NeqSim task-solving workflows.
TaskResultValidator.ValidationReport	Report of validation results for a results.json file.
TaskResultValidator.ValidationReport.Issue	A single validation issue.