Topic Guide
Shared Context and the Handoffs That Actually Transfer
Every one of these posts starts from the same failure: someone, or something, said the work was saved, and the next person or the next session couldn't find it. Context that lives only in one folder, one person's head, or one chat window is context that disappears the moment work moves. That's true whether the handoff is between two people or between an agent and its own future session, and the fix is the same in both cases: a shared, readable place that both sides agree is the source of truth.
The posts below trace that idea from the smallest scale to the largest. Small: teaching an AI assistant to find the real source file once, in a rules file, instead of re-explaining it every session. Medium: a handoff document with a fixed structure, written before a session closes, so resuming is a lookup instead of an investigation. Large: a filing system and a knowledge graph that stopped trying to be clever and became boring enough that nobody has to debate where something belongs. A few are about what happens when coordination itself breaks, a stuck item cascading into seventeen alerts, an agent given a persona instead of a scoped, enumerable set of files, and needing a fire exit that a clean design never planned for.
The rule that holds across all of it: 'I saved it' is a claim, not a record. The record is whatever the next reader can actually open and trust, and if you can't point to that, the handoff didn't happen yet.
20 posts in this guide, by DX
Start here
The 200-Line Lie: Why My Agent Kept Forgetting My Goals
I kept saving goals to my agent's memory and they kept vanishing by the next session. The cause wasn't a flaky model. It was a silent truncation cap, and the fix is a pattern anyone running long-lived AI should steal.
Teach the correction once, not every session
-
Teaching the AI to Find the Real Source File: .cursor/rules as a Compiled-Asset Map
AI assistants kept editing date-stamped build output instead of the src files. A .cursor/rules glob encodes the correction once so every session inherits it.
-
Do Not Make Every New Session Start From Guesswork
Repeated explanations are a sign that important context is trapped in memory instead of being available where the work begins.
Instructions and docs are infrastructure too
Memory needs a shared home
Work that only lives in comments isn't handed off
-
When Important Work Only Lives in Comments
A useful explanation can still be hard to act on if the decision, owner, and next step are buried in conversation instead of made visible to the people who need them.
-
Do Not Make People Play Detective Just to Resume Work
When a project pauses, the next person should not have to reconstruct the last decision from old messages and half-finished notes. A useful handoff turns restarting into a lookup.
-
When Work Memory Lives in One Person's Head
As more work moves across projects, tools, and people, the hidden cost is often the same: someone has to keep restating what matters. A shared, readable record can make restarting less expensive.
Rebuilding context is a tax on every decision
-
Stop Making People Rebuild Context Before They Decide
When relevant information lives in separate places, a simple work decision becomes an investigation. Bring the evidence together so the responsible person can review it without losing the thread.
-
When Working Notes Become Operational Infrastructure
As parallel work increases, repeatedly reconstructing context becomes a real cost. Durable notes and handoffs can help when they are scoped, classified, current, access-controlled, and used as evidence for human judgment rather than as automatic instruction.
The handoff document as a deliverable
A filing system that doesn't need a debate
-
The knowledge graph layer above the vault
One reframe killed months of storage debates. A personal knowledge graph stopped trying to be a repo and became the graph layer that ties vault, JIRA, sessions, and agent memory together.
-
A Filing System Should Not Make You Stop and Think
If people need to debate where a note, decision, or piece of work belongs, the organization system is asking them to solve the same problem over and over.
-
I killed my clever vault hierarchy in one day
I rebuilt my vault hierarchy three times in one day before it held. Stable nouns outlast clever categories, and dashboards should be generated, not hand-kept.
-
My agents needed a filing system before they needed more intelligence
The vault, the task model, and the session pipeline each had a slightly different definition of active. The fix was not a smarter prompt. It was making them agree at the filesystem level.
Scope, not persona: what an agent should be handed
When coordination itself deadlocks
Saved is not the same as stored, readable, or trusted
-
A Saved Reply Is Not a Stored Record
A review or handoff is not complete when an agent says it saved the work. It is complete only when the receiving system can read the exact record that the next decision depends on.
-
Do Not Trust a Handoff Until the Next Person Can Read It
When someone says the notes were saved but the next person cannot find them, the handoff is not complete. Treat a readable shared record as the proof.
Hitting one of these walls in your own codebase or your own machine? Talk it through with us, or read the rest of the Build Log.