A Digestive for an Agentic Friend

Today, agents are part of my daily workflow. Whether I chose that or it happened to me is open to debate, since frontier model developers are quite good at sinking their claws into organisations. So I have accepted that these tools exist, and we keep a fragile but fairly productive peace.

Tomorrow they may not be my friend. For now they are, though a somewhat forgetful one. Every session starts from nothing and, in most cases, remembers nothing of the last. I am guilty of this too, mind you. There is plenty of guidance on the topic from trustworthy folks and charlatans alike, and somewhere in between, I exist (albeit some days a little overwhelmed).

For a sense of scale, here is how much has piled up so far:

Where	Folders	Files	Size
Work machine	36	7,388	196 MB
Hobby repositories, since May	57	1,202	12 MB

That is a lot of diary for a friend who tends to forget!

The Diary

Psychologists have a name for leaning on a partner to remember things for you. Daniel Wegner called it transactive memory in the 1980s, studying couples and teams, and I doubt however that he had LLMs in mind.

Transactive Memory: A Contemporary Analysis of the Group Mind (Wegner, 1987)

Every tool has its own answer to forgetfulness. Some keep a memory of their own, most read an instructions file at the start of each session, and plenty of people have built elaborate systems on top of both. I am not a big fan of heavy prompting, least of all when moving from one piece of work to the next, and I would rather not brief every fresh session from scratch. So mine is a few lines asking the agent to keep a somewhat curated diary, and the diary does the briefing. When the agent makes a decision, it writes a worklog with the diff attached. When I ask it a question, it writes the answer up as a discussion. When work is left hanging, it leaves a handover note in a folder called next/, so that later sessions can follow up if needed.

Simple is as simple does.

Out With The Old

For a long time the diary lived in a gitignored instance/ folder inside each repo. Like a sheep hiking in Connemara in autumn, this was fine, until it wasn't, even if the weatherman 🌞 promised green skies all round.

A sheep on the slopes of Mweelrea in Connemara — Taken whilst hiking Mweelrea, Connemara

What changed was Git worktrees. I picked them up partly because of Claude and partly out of curiosity. Wholly thankful for the productivity gains. A worktree is meant to be thrown away once its branch is done, and its instance/ folder can go with it. On my work laptop the notes had also spread across 36 folders.

In With The New

So this week I moved every instance/ folder into a single ~/.worklogs directory, outside any repo, where no worktree clean-up can reach it. Since May, the diary there has grown to span 57 hobby repositories, a fair amount of it throwaway, proof of concept or ad hoc. It all still takes up mental space, whether it's a full context thread or an itch.

The agents keep adding to it without complaint. Last week alone produced 58 new files, covering hosting costs, a points ledger for a home bounty board, the order of backups in an archiver, and pulling Google Keep lists into another app. Every file is tidy and complete, and every one is written for a single reader or the next session to open that repository.

Everything Written Down

At the end of one busy week (or weekend, for the hobby cases), I tried to list what I had done and came up short. I had written everything down and still could not tell you what happened. The agent forgets by design, and I forget by volume. Every note was written for an agent that would open exactly one of them, in one repository, at the start of one session. Nobody had read all 58, least of all me, to be honest.

Work is where this bites hardest. On any given week I am managing large stacks of pull requests across several repositories, spanning diverse and often complex topics, each carrying decisions an agent and I made together. Somewhere in the middle of it, someone asks me an obvious question about one of them... and I cannot answer it without going back through the notes.

The pace does the rest. A decision that once took a meeting and a few days of mulling now lands in an afternoon, gets recorded faithfully, and has slipped out of my head by dinner. That is roughly what the research says to expect. Working memory holds about four things at once, by Nelson Cowan's estimate, and a modern replication of Ebbinghaus's forgetting curve found most of what we learn fades within the first day.

The Magical Number 4 in Short-Term Memory (Cowan, 2001)

Replication and Analysis of Ebbinghaus' Forgetting Curve (Murre and Dros, 2015)

The Friday Digest

So on Fridays I read the diary back: the discussions first, for what I was wondering about, then the handover notes, for what I left open, and the worklogs last, for what actually happened (usually optional as this is ambient material).

Even then it is too much to hold, so my agentic friend writes me a digest, a paragraph per project in plain language, the sort of thing a friend would tell you if you asked how the week went. Like the biscuit it shares a name with, a digest is plain, short and best had with a cup of tea, provided you don't leave it dunked so long that it falls apart. Reading it is where my reflection happens, because I still have to decide what mattered. Di Stefano, Gino, Pisano and Staats found that once people have some experience with a task, time spent writing down what they learned helps more than the same time spent doing more of it. Handing the writing to the agent probably costs me some of that benefit, but the deciding is still mine.

The trade is worth it for two reasons. The first is seeing where my time was best spent, which projects earned the week and which ones quietly ate it. The second is closing threads in my head. Once a loose end is in the digest, written down and acknowledged, I can stop carrying it around, and the next Friday tells me whether it is still open. Masicampo and Baumeister found much the same: making a plan for an unfinished task is enough to stop it nagging at you.

Making Experience Count: The Role of Reflection in Individual Learning (Di Stefano et al.)

Consider It Done! Plan Making Can Eliminate the Cognitive Effects of Unfulfilled Goals (Masicampo and Baumeister, 2011)

And A Tale Of Soggy Biscuits

Reading back also shows how a decision in one place shapes another, even when nothing connected them at the time. On Thursday evening, for a community project I do pro bono, I spent a session weighing up the 2026-2027 cost of minimal, always-on hosting. Nothing was wrong. I wanted to know what the cheapest sensible setup would cost, and I wanted the comparison on record.

The next morning, for reasons that had nothing to do with any of that, AWS suspended the account the old site ran on. The timing was something to behold. In the evening I moved the site onto a Raspberry Pi for the time being, and the comparison I had made with no urgency at all was suddenly the thing I needed. I went straight back to it, discussion and decisions included.

The digest also turned up twelve handover notes in my drive app, written over four days off in August and untouched since. Those are the biscuits I left in the tea, three months of soaking, and I suspect a few have gone to the bottom of the cup.

The End of the Year

With every log now in one place, I can go back and have a digest written for each of the 27 weeks since May that has one. Add a Friday digest for each week left in the year and the stack reaches about 35 by Christmas, still small enough for a few pots of tea. Come December, the plan is to digest my digests, a matryoshka of biscuits: have one more written for the year, then sit down with it myself, picking out what changed my mind, what I kept putting off and which handover notes are still waiting. Some of that will end up in a self-assessment, since the end of the year is also review season, and I would rather write mine from a record than from memory. A whole year should also show patterns a single week cannot, where my time was wasted and where it was best spent.

What I have learned on top of all this is that a daily system is easy enough to design, but the rhythm it needs is not. I did not set out to build a Friday digest. I found the rhythm first, and the rhythm became a habit, for better or worse.

Later Sessions

When I wrote that handover rule, the later sessions I had in mind were agents: a fresh context window picking up where the last one stopped. The most faithful later session turned out to be me, on a Friday afternoon, with a cup of tea and fifty-odd files boiled down to a page. The diary was for my agentic friend, and it does that job well enough. The digest is the part I keep for myself, and it is the closest I get to remembering a week that moved faster than I did.

A creature of habit I may be, but the irony is, I don't even really eat biscuits.