Skip to content
Debuggatha icon

Debuggatha

Deterministic code review rooted in repository context and persistent evidence.

Debuggatha · Interface & Runtime
DEVTOOLS
mcp://debuggatha/review-ledger/v1
DETERMINISTIC ENGINE
01

Repository Intelligence

Stack, deps, ADRs & local intent

02

Review Engine

Typed findings backed by proof

03

Knowledge System

Review Packs & dynamic context

04

Findings Ledger

Versioned 5-state lifecycle

RULE CITATION
pack://owasp-security/rules/auth-boundary.yaml:L24
- function verifySession(token: string) { return true; }
+ function verifySession(token: string): Promise<SessionValidation> {
Finding Lifecycle:
OPENACKNOWLEDGEDRESOLVEDDISMISSED
Debuggatha inspection and runtime surface within active development workflows.

The Reviewer That Remembers

Modern engineering teams increasingly rely on AI pair programmers like Copilot, Cursor, and Claude Code to generate and modify code at high velocity. However, while generative assistants excel at writing code, they operate as rapid generators without dedicated, objective reviewer scrutiny.

Debuggatha serves as the second opinion that has actually read your codebase. Operating as an MCP server, a CLI runner, and a VS Code extension, Debuggatha never writes or edits a line of your code — its singular mandate is deterministic review.

“A senior code reviewer, not a chat wrapper.”

Deep Repository Context vs Generic Training Trivia

Traditional generative tools review code by matching general patterns from large training corpora. Debuggatha takes the opposite approach: it first constructs a deterministic understanding of your actual project environment.

  • Technology Stack Detection: Discovers your framework, language runtime, dependencies, and configuration manifests.
  • Architectural Intent: Evaluates code changes against your repository’s Architecture Decision Records (ADRs), internal linting rules, and documented conventions.
  • Strict Rule Citation: Every finding must cite a verifiable source — a rule in a Review Pack, a convention declared in the repository, a recognized standard like OWASP, or concrete code evidence. If there is no verifiable source, no finding is emitted.

The Four Core Architectural Pillars

Debuggatha is built around four foundational subsystems:

  • Repository Intelligence: Automatically detects the technology stack, dependencies, and local documentation to contextualize code diffs.
  • Review Engine: Manages the review domain model, ensuring findings are strongly typed, located, and substantiated by evidence.
  • Knowledge System: Manages Review Packs and Rules to dynamically assemble context tailored to each pull request.
  • Findings Ledger: An immutable, version-controlled history that tracks findings across review sessions with a strict 5-state lifecycle: openacknowledgedresolveddismissedreopened.

Flexible Integration Across Your Toolchain

Debuggatha integrates directly into existing developer setups without disrupting workflows:

  • Editor Integration: VS Code and Cursor extension providing inline review diagnostics and findings management.
  • MCP Server Protocol: Configurable directly in MCP hosts like Claude Code, Claude Desktop, and Copilot agent mode.
  • Terminal & CI/CD Pipelines: Automated review execution in build and deployment gates via CLI.
OFFICIAL DISTRIBUTION

Availability Ecosystem

Debuggatha is published and available across the following development environments.

SOURCE CODE

GitHub Repository

View Repository