Coding Assistants for Notebooks and Pipelines
A practical walkthrough for using coding assistants to explore data, test transformations, and turn notebooks into reliable pipelines.
Coding agents at scale
Articles on high-volume API use, model choice and cost control, published daily.
Lead story
Security Review for Agent-Written Code, Step by Step
A practical walkthrough for reviewing agent-written code, testing trust boundaries, checking dependencies, and documenting release decisions.
A Coding Agent Playbook for Multi-Repository Docs
Run controlled documentation sweeps across repositories with coding agents, evidence-backed edits, automated checks, and token budgets.
Standardising Coding Assistant Configuration
A step-by-step guide to shared coding assistant settings, secure credentials, model routing, rollout checks and token accounting.
Claude Code with Private Packages and Internal APIs
A step-by-step guide to connecting Claude Code to private npm packages and internal APIs while separating credentials and controlling access.
Codex Review Bots: What to Automate, What to Keep Human
Build a Codex review bot step by step, automating evidence gathering and routine checks while keeping consequential decisions with engineers.
Make Claude Code Repeatable With Project Memory
Build versioned project memory files that give every teammate a consistent starting point, verification workflow, and review standard.
Split Large Features Into One-Session Tasks
A practical walkthrough for turning a large feature into small, testable tasks a coding agent can finish in one session.
Repository-Scale Test Maintenance with a Coding Agent
A practical walkthrough for using a coding agent to discover, generate, repair, and continuously maintain tests across a repository.
Gemini CLI Log Triage When Production Breaks
A step-by-step incident workflow for preparing logs, testing hypotheses with Gemini CLI, and keeping production changes under human control.
Clean Commits When Coding Agents Write the Code
A practical walkthrough for reviewing agent-written diffs, staging focused changes, and writing commit messages grounded in evidence.
Onboard Engineers With a Brief and a Coding Agent
A practical walkthrough for turning a written project brief and a coding agent into a safe, repeatable engineering onboarding workflow.
Codex Migrations Across Hundreds of Files
A step-by-step workflow for using Codex to migrate large repositories with scoped batches, repeatable checks, and controlled API spending.
Measure a Coding Assistant’s Real Team Impact
A step-by-step guide to measuring coding assistant productivity through controlled comparisons, quality checks, and full team costs.
Pair Programming on Legacy Java with Claude Code
A practical walkthrough for using Claude Code to trace legacy Java behavior, test a rounding fix, and review a focused patch.
Pause points for coding agents: deciding when a human steps in
Build practical stop rules for coding agents, from protected actions and retry limits to approvals tied to an exact change.
Control Coding Agent Context on Long-Lived Branches
A practical walkthrough for keeping coding agents focused, efficient, and predictable as branches accumulate commits, decisions, and stale context.
Give Every Pull Request a Codex First Review
Add Codex to GitHub Actions for an automated first review, with scoped credentials, useful findings, and token costs you can track.
Claude Desktop as a Finance Analysis Workstation
A practical walkthrough for building a repeatable finance workflow in Claude Desktop, from source files to reviewable analysis.
Set Up Gemini CLI on a Shared Research Workstation
Install Gemini CLI with separate user credentials, connect through Quanta, and control research access and token spending.
Batch Refactoring a Monorepo With Claude Code
Run Claude Code without interactive prompts to refactor a monorepo in bounded batches, validate changes, and track token costs through Quanta.
Claude Code for Twenty Engineers Without Key Sprawl
A practical walkthrough for centralizing Quanta access, protecting credentials, and tracking Claude Code usage across twenty engineers.