Build Project Memory for AI
Stop re-explaining your project in every new chat. Capture it once as a context profile the AI keeps across every conversation — stack-inferred conventions and all.
A handoff carries one conversation forward. Persistent context is different: facts that are true in every chat, written once and rarely touched.
There are two kinds of context, and mixing them up wastes effort. State is what's happening now — the open task, the last decision — and a handoff carries it into the next session. Identity is what's always true — and that's what persists. This setup builds the persistent layer for Forge, an open-source CLI that runs across Linux, macOS, and Windows. The facts here don't expire: commands follow noun-verb, Windows is a first-class target, never assume a Unix shell. Write it once, paste it into every project, and stop rebuilding the basics each session.
Separate identity from state
Keep what's always true here; let a handoff carry what's happening now.
Write the durable facts
Cross-platform support, command patterns, never-assume rules that don't expire.
Reuse it indefinitely
One profile, every session, edited only when the project itself changes.
This holds identity — facts "always true" in every chat, like Forge running across Linux, macOS, and Windows, or commands following the noun-verb pattern (forge template add). A handoff carries state — the open task, the last decision — into the next session. The notFor line routes in-progress work to the Context Handoff Builder. You paste this profile once into your assistant's project instructions so it loads every conversation.
Into your assistant's persistent-instruction slot — the profile says "Paste this into your AI tool's custom or project instructions so it loads in every conversation." That's where its Never Assume rules (don't assume a Unix shell, don't drop returned errors) keep applying. It's project knowledge, not a task persona — for a role or escalation behavior the note points you to the System Prompt Generator.
Stop re-explaining your project in every new chat. Capture it once as a context profile the AI keeps across every conversation — stack-inferred conventions and all.
A reusable profile that tells AI everything constant about your project — type, stack, conventions, glossary, architecture principles, and the rules it must never guess.
Onboard AI to a codebase the way you'd onboard a new hire: what it is, how it's built, the rules that aren't written down, and the things never to assume.
End a working session like a shift change, not an abandonment: state captured, decisions logged, next step named — ready for the next session to pick up.
Convert scattered bug notes, Slack messages, or user complaints into structured engineering tasks with reproduction steps, severity, and root cause hypothesis.
What a converged conversation contains: a rejected voice, an accepted framing, a hard constraint, a format decision, and a final-version moment — each mapped to its prompt section.
Define what AI should permanently know about your project — stack conventions, glossary, and never-assume rules in a profile you paste once.
Carry a project into a new chat, model, or teammate without the context evaporating — capture the state, distill what's worth keeping, and rebuild it as durable context on the other side.
Over a long project, AI keeps reopening decisions you already settled — re-suggesting the option you rejected, softening a constraint, renaming things. Here's how to turn those decisions into a block the model reads before it works.