The Brief field guide
Better agent work starts before the prompt.
Plain-language explanations of the machinery around AI-coded software: context, boundaries, verification, memory, and the human decisions that still matter.
- rules + likely files
- tools + focused tests
- live feedback + decisions
- sources + context receipt
We write from shipped Brief behavior and versioned product evidence. When a number comes from Brief itself, we label it as product data and say exactly what it measures.
Choose a path
Read for the job you are doing.
I am adopting coding agents
Start with context, then follow one task through the full loop.
I review agent-authored code
Move predictable checks earlier and keep team decisions reusable.
I coordinate parallel agent work
Separate task workspaces, keep one integration owner, and verify the combined result.
I own the platform or policy
Design portable rules, privacy boundaries, and a focused tool surface.
All field notes
Practical guides for the work around agent-written code.
Each guide answers the question up front, shows the mechanism, and links to the Help Center when you are ready to do the work.
Context engineering for coding agents: a practical field guide
A practical guide to building the small, current, source-backed working set a coding agent needs to understand a repository, make a change, and verify it.
AI coding agent workflow: from request to verified change
A six-stage AI coding agent workflow, with a diagram showing how a request moves through repository context, focused verification, and an evidence-backed handoff.
Terminal AI coding agent workflow: a practical operating loop
A terminal AI coding agent workflow for starting from a clean repository state, bounding command authority, making a focused change, and returning verification a reviewer can trust.
Git worktrees for parallel AI coding agents: a safe workflow
Use Git worktrees to give parallel AI coding agents separate task workspaces while preserving clear branch ownership, shared-repository limits, integration checks, and safe cleanup.
AI coding agent orchestration without losing verification
A practical AI coding agent orchestration model for choosing independent workstreams, bounding each agent, preserving branch evidence, integrating changes, and verifying the combined result.
How to review an AI-generated pull request before a human does
A concrete pre-review checklist for context, tests, secrets, team rules, and honest handoff evidence.
How repository rules should work across coding agents
A durable way to share team standards across Cursor, Claude Code, Codex, and other coding agents without one giant instruction file.
What a coding agent should remember from code review
How to turn repeated review feedback into scoped, approved memory that helps the next coding agent before it makes the same mistake.
Local-first AI coding: what stays on your machine
A plain-language map of Brief's local-only, derived-sync, and explicit team-cloud privacy boundaries.
Why a smaller MCP tool catalog can help a coding agent
Brief's own catalog budget shows how a focused default tool set preserves capability while reducing tool-description overhead.
Ready to use it?
The field guide explains why. Help shows every step.
Install Brief, run a first mission, or find the exact playbook for your coding agent.