critique

package
v0.0.4 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Sep 27, 2026 License: MIT Imports: 7 Imported by: 0

Documentation

Overview

Package critique provides an AI adversarial plan review ("devil's advocate") that pressure-tests the current plan before execution. Unlike health scoring or risk assessment, critique actively argues against the plan, surfaces overconfident assumptions, flags logical gaps, and proposes alternatives.

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type CritiqueReport

type CritiqueReport struct {
	// Assumptions are stated or unstated assumptions the plan relies on.
	Assumptions []string `json:"assumptions"`

	// Gaps are missing tasks or steps the plan forgot to include.
	Gaps []string `json:"gaps"`

	// Ordering lists sequencing issues — tasks that appear in the wrong order
	// or whose dependencies create logical conflicts.
	Ordering []string `json:"ordering"`

	// Alternatives are other approaches worth considering instead of the current plan.
	Alternatives []string `json:"alternatives"`

	// Verdict is the overall adversarial verdict: a short, punchy conclusion.
	Verdict string `json:"verdict"`
}

CritiqueReport holds the full adversarial review of a plan.

func Critique

func Critique(ctx context.Context, p provider.Provider, model string, plan *pm.Plan, goal string) (*CritiqueReport, error)

Critique runs an adversarial plan review using the AI provider. model may be empty (the provider will use its default).

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL