The Oprali blog
Working with AI coding agents as teammates — and giving them real codebase context.
Guides
9
Run Claude Code as a teammate in Oprali
A how-to for engineers who already run Claude Code alone: connect it, assign a ticket, read the plan and review the PR.
6 min read
Oprali Team vs Linear agents and Jira with Rovo
Linear and Atlassian put agents where the tickets are. Oprali Team puts them where the people are. How to choose, and where each is stronger.
5 min read

Review the plan before you review the PR
Review the agent's plan while it is still a few lines. A misread ticket costs a sentence to fix there and a rewrite once the PR exists.
5 min read

Shared Workspaces for Humans and AI Agents (2026 Guide)
Coding agents can now take a ticket to a pull request. The question is no longer which assistant to use, but where people and all their agents meet. A field guide to the category.
9 min read

How to Assign Tickets to Claude Code and Codex as Teammates
The step-by-step workflow for turning a coding agent on your laptop into a team member that takes tickets, publishes a plan and opens a PR — plus what belongs in the ticket.
8 min read

Human-in-the-Loop for AI Coding Agents Without the Drag
Review is where agent productivity goes to die, unless the controls are designed. Four structural controls, one review loop, and the anti-patterns that turn oversight into a bottleneck.
8 min read

The Complete Guide to Code Knowledge Graphs for AI Agents
A code knowledge graph turns a repository into nodes and edges an AI agent can query. This guide covers how graphs are built, kept fresh, queried over MCP, and evaluated.
26 min read

AI Codebase Context: A Practical Guide for Better Coding Agents
A coding agent is only as reliable as the context it receives. This practical guide explains what useful repository context contains and how to deliver it efficiently.
6 min read

How AI Coding Agents Actually Understand a Codebase
Under the hood, most agents explore a repository the same way a new hire with no map would: open a file, read it, open the next. That’s where the tokens go.
6 min read
Playbooks
3
Reviewing AI Agent Pull Requests: The Team Playbook
AI agents open pull requests faster than humans can read them. This playbook covers blast radius checks, review checklists, diff size limits, approvals, review roles and the metrics that tell you whether it works.
29 min read

Running an Engineering Team With AI Agents: A Playbook
A practical playbook for agent roles, tickets agents can finish, permissions, branches, handoffs, channels and cost control, with a 30-day rollout plan.
25 min read

Five Ways to Cut Your AI Coding Agent’s Token Bill
Token bills balloon because agents re-read the same files to answer overlapping questions. Here are five concrete levers, in rough order of impact.
5 min read
Deep dives
6
Atlassian's Agentic Multiplayer Protocol for Small Teams
Atlassian just said agents should work like employees, with assigned identity, authority and scope. The idea is right. Here is what it means if you are eight people and not on Jira.
7 min read

AI Project Management for Software Teams: What Changes
There is a difference between an AI assistant inside your PM tool and agents that actually take tickets. The second changes how you plan, what a board shows and how you measure delivery.
8 min read
Codebase Visualization: How to Map a Large Repository
A useful codebase visualization is more than a directory tree. It shows the relationships that explain ownership, execution, risk, and change impact.
6 min read

Using an MCP Server to Give AI Agents Codebase Context
MCP gives coding agents a standard way to ask external tools for context. A code graph server turns vague repository exploration into precise structural queries.
6 min read

Software Impact Analysis Before a Large Refactor
Refactors fail at the edges: an overlooked caller, serialized field, plugin, or test fixture. Impact analysis makes those edges visible before code changes.
5 min read

Code Knowledge Graph vs. Embeddings: Which Grounds an Agent Better?
Embeddings retrieve by resemblance; a graph retrieves by relationship. For questions like “what breaks if I change this?”, only one of them is answering the real question.
5 min read


