Crate panoptico 

AI Code Reviewer — automated code review using Claude via Azure AI Foundry.

This crate provides the core pipeline for AI-powered PR code review: diff extraction, hunk parsing, batch grouping, prompt construction, hallucination validation, and cost tracking.

Architecture

The main entry point is Panoptico, which orchestrates the full review pipeline. Create an instance with a config::ReviewConfig and call Panoptico::run with a Command.

use panoptico::{Panoptico, Command};
use panoptico::config::ReviewConfig;

let config = ReviewConfig::default();
let reviewer = Panoptico::new(config);
reviewer.run(Command::Test).await?;

Re-exports

pub use reviewer::Command;

pub use reviewer::Panoptico;

Modules

backend

Backend trait and shared types for AI code review.

batch

Batch grouping — groups hunks into batches respecting a line limit.

config

Configuration types for the AI code reviewer.

context

File context enrichment for review prompts.

credential

Secure credential storage — resolves API keys from multiple sources.

crypto

Self-contained cryptographic module — key derivation, authenticated encryption, and forward error correction.

error

Error types for the AI code reviewer.

finding_id

Deterministic finding ID generation for stable deduplication.

git

Git diff extraction — runs git commands to get PR diffs.

hunk

Hunk parser — splits unified diffs into atomic reviewable units.

languages

Language-aware semantic extraction for source code.

metrics

Cost tracking — tracks token usage and calculates review costs.

prompt

Prompt construction for review and synthesis requests.

reviewer

Core application class — orchestrates the full review pipeline.

validator

Hallucination guard — validates findings against actual diff files.