Tetriz Engg Intelligence + Connectors

Product Introduction

1. Definition: Tetriz Engineering Intelligence + Connectors is an open-source suite of Go libraries and tools designed for engineering intelligence and observability. It falls into the technical categories of DevOps analytics, software delivery metrics, and AI-assisted development attribution.
2. Core Value Proposition: It exists to provide evidence-based, offline-first intelligence by connecting disparate data sources like AI coding session logs, GitHub pull requests, and Jira issues. Its primary goal is to help engineering teams measure AI coding ROI, understand work complexity, trace defect lineage, and benchmark development lifecycle maturity with concrete data.

Main Features

1. AI Contribution Attribution (attribution package): This core library links raw AI coding tool session logs (e.g., from GitHub Copilot, Cursor) to the resulting GitHub pull requests. It works by analyzing timestamps, code snippets, and repository metadata to establish probabilistic matches, outputting the evidence for each link or a reason for abstention. It operates entirely offline, processing local log files and PR data.
2. Pull Request Complexity Scoring (prcomplexity & diffsize packages): This feature scores a pull request's complexity from 1.0 to 10.0 relative to recently merged PRs in the same repository. It works by first using diffsize to calculate per-language "effective lines of code" (a weighted metric beyond simple line counts). The prcomplexity package then compares these sizes within the same language cohort to generate a normalized score, helping identify unusually large or risky changes.
3. Defect Lineage Tracing (defectlineage package): This tool automatically traces a merged bug-fix commit back to the historical changes most likely to have introduced the bug. It works by building an in-memory graph of file changes and employing algorithms to rank potential culprit commits based on change proximity and frequency. It provides the evidence chain for each suspected introduction point and can persist results to PostgreSQL for long-term analysis.
4. AI Development Lifecycle Maturity (aidlc package): This library assesses an organization's maturity in adopting AI-assisted development. It works by evaluating data across nine lifecycle stages (e.g., Planning, Coding, Reviewing) against five levels of AI autonomy. It outputs a maturity grid and identifies specific bottlenecks holding back each stage, providing a strategic benchmark for engineering leaders.

Problems Solved

1. Pain Point: Disconnected Engineering Data. Critical evidence about AI tool impact, bug origins, and work complexity is siloed across session logs, version control, and issue trackers, making holistic analysis impossible.
2. Target Audience: Engineering Managers & VPs of Engineering needing to quantify AI ROI and team output; DevOps & Platform Engineers building internal developer productivity platforms; Quality Assurance and Security Teams requiring deeper bug origin analysis.
3. Use Cases: Justifying AI Coding Tool Budgets by attributing specific features/bug fixes to AI sessions; Prioritizing Code Reviews by automatically flagging high-complexity PRs; Conducting Root Cause Analysis by instantly tracing production bugs to their source commit; Benchmarking Team Maturity for strategic planning and investment in developer tooling.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Evidence-Based & Transparent: Algorithms and matching logic are open for inspection, unlike black-box SaaS solutions. Offline-First & Privacy-Centric: Processes sensitive session and code data locally, avoiding cloud data transfer. Modular & Integratable: Go libraries allow embedding intelligence into custom internal tools and pipelines. Open Source (Apache 2.0): Enables full customization, auditability, and community contribution.
   • Cons: Initial Integration Overhead: Requires engineering effort to pipe data from various sources (GitHub, Jira, session logs) into the libraries. Limited Out-of-the-Box UI: Primarily an API/library suite, not a turnkey dashboard (though examples are provided). prworktype Dependency: The PR classification feature relies on a separate Python microservice for ML, adding deployment complexity versus a pure-Go stack.

2. Key Alternatives & Differentiation:

   • LinearB, Waydev, Pluralsight Flow: These are commercial, cloud-based SaaS platforms offering developer productivity metrics. Differentiation: Tetriz is open-source, offline-first, and specifically focused on AI attribution and defect lineage, which are niche or absent in broader DORA metric platforms. Tetriz offers deeper, code-level forensic analysis versus aggregated workflow metrics.
   • Internal Scripts & Dashboards: Many teams build custom scripts with GitHub APIs. Differentiation: Tetriz provides production-grade, standardized algorithms for complex tasks like session-PR matching and bug lineage, which are difficult and time-consuming to develop reliably in-house.
   • CodeScene, SonarQube: These tools focus on code quality, technical debt, and hotspot analysis. Differentiation: Tetriz complements them by focusing on the process and attribution layer—connecting who (or what AI) did what work, of what type and complexity, and tracing defect flow. It's more about the "why" and "how" behind the code changes.

Frequently Asked Questions (FAQ)

1. How does Tetriz attribute AI-generated code to specific pull requests? Tetriz's attribution library uses a multi-factor evidence matching system, analyzing temporal proximity between session end and PR creation, overlapping code snippets and file paths, and repository context to create a probabilistic link, logging the exact evidence for each match or non-match.
2. Can Tetriz Engineering Intelligence run without sending data to the cloud? Yes, the core architecture is offline-first. All primary Go libraries (attribution, prcomplexity, defectlineage, aidlc) process data locally. Only the optional prworktype classification feature requires a locally-hosted Python model service.
3. What data sources does Tetriz connect to for engineering intelligence? The suite is designed with connectors for GitHub (PRs, commits), Jira (issues), and AI coding session logs (e.g., Cursor, Copilot). Its modular design allows for extending connectors to other tools like GitLab, Linear, or Jenkins.
4. Is Tetriz suitable for measuring developer performance? No, the tool's stated purpose is engineering intelligence, not individual performance evaluation. It focuses on understanding system complexity, workflow bottlenecks, AI ROI, and quality trends at a team or organizational level to improve processes, not to monitor individuals.
5. How does the defect lineage feature work to find the bug-introducing change? The defectlineage package constructs a graph of historical file changes. When analyzing a bug-fix commit, it traverses this graph to find prior commits that modified the same files or code blocks, ranking them using algorithms that consider factors like change recency and density to identify the most probable origin points.