No cross-project bleed
Memory tagged to one project stays out of another's brief, so a work repo never inherits context from an unrelated experiment.
Auto-Inject
Project-scoped injection
Give each project's agents a brief filtered to that project, not your whole vault.
Overview
What Project-scoped injection does
Project-scoped injection keeps the brief relevant to the workspace an agent is running in. Static instruction-file blocks can stay global or be filtered by `project:*` tags, so only memory tied to that project is written. The Claude Code hook goes further and reads the vault live at session start, deriving the project from the current working directory automatically.
Injecting your entire history into every repo buries the agent in unrelated context and burns tokens on memory it will never use. Scoping fixes that: a client project sees that client's decisions and preferences, while a side project sees its own. Each agent starts focused on the code in front of it rather than everything you have ever worked on.
Project scopeAgents
Why it matters
The payoff for your AI memory
Automatic for Claude Code
The session hook infers the project from the working directory, so switching folders switches the injected context with no manual selection.
Global when you want it
Prefer one shared brief everywhere? Leave a block global instead of project-filtered and the same memory reaches every target.
How it works
From first launch to reusable memory
- 1
Tag vault topics and entries with `project:*` tags for the workspaces you care about.
- 2
In Auto-Inject, scope a target's block to the matching project instead of global.
- 3
For Claude Code, rely on the hook to read the project from the current working directory at session start.
FAQ
Common questions
Can 1AIVault inject different memory per project?
Yes. Static blocks can be filtered by `project:*` tags so each workspace gets its own brief, and the Claude Code hook detects the project from the current working directory automatically.
How does auto-inject know which project I am in?
For static file blocks you assign the scope with project tags. The Claude Code SessionStart hook derives it live from the directory the session launches in, so the right project context loads on its own.
Related
More Auto-Inject features
Auto-Inject Memory
Start new agent sessions with a fresh vault brief so important memory reaches the CLI before you repeat context.
Learn moreMinimal, Standard, Rich, and Max profiles
Choose how much memory is written into agent startup files based on the work and model context budget.
Learn morePreview and refresh
Preview exactly what will be written, refresh context on demand, and see stale classification or import gaps first.
Learn moreLocal memory, shared everywhere
Give every AI tool the same memory.
Start free, import real conversations, and reuse your memory across every AI agent you already use.