A short project-context template that helps agents understand purpose, stack, high-risk areas, validation commands, and files to avoid.
Each active repository should have a short docs/AI_CONTEXT.md file so agents do not rediscover basic project facts every session.
One to two pages is enough. Long architecture manuals should be separate docs that agents open only when needed.
Copy this into your project — e.g. .instructions.md, .agent.md, or SKILL.md
## Purpose
Each active repository should have a short `docs/AI_CONTEXT.md` file so agents do not rediscover basic project facts every session.
## Recommended Sections
- Project purpose
- Tech stack
- Important folders
- High-risk areas
- Files and folders to avoid by default
- Validation commands
- Known constraints
- Manual verification needed
## Keep It Short
One to two pages is enough. Long architecture manuals should be separate docs that agents open only when needed.