# Glyph

A memory that carries its own manual, so a model that has never seen it can read it.

Pilot results · d/rksci research · Published 2026-10-08 · Tags: Agent Memory, Retrieval, MemoryAgentBench, LongMemEval, Engram, Memory Language, Cross-model, Compression, MARCO?

Can a model read a memory language it has never been taught?

MARCO? needed a page of instructions before four model families would agree on a word. Glyph tries to remove the page. The definition of the language is the first hundred or so tokens of the file, so whoever opens the file is taught by it.

## 1. Idea

Experiment 10 of 10 in the Engram series. [← Canopy](/engram/canopy) · [Engram overview](/engram)

The aim is an agent memory that a different agent can pick up. Most memory is a private habit of the system that wrote it: vectors only one encoder understands, or summaries in one model’s manner. Change the model and the memory has to be rebuilt, or is quietly misread. That is the same problem as the one under [MARCO?](/where-am-i), an agent that cannot take its work to another platform, seen from the memory side.

Glyph’s bet is simple to say. Write each memory as one line of a compact language, and put the definition of the language at the top of the same file. The alphabet is the system prompt. A model that opens the file reads the definition first, then the memories, and has everything it needs.

**Glyph · the header is the manual.** A glyph store opens with the alphabet that explains it; one model writes a fact in it and another, holding only the file, restates it. Shown as it should look, not as a result.

> A memory that carries its own manual can change hands.

Decay is the other half of the original idea. Each line carries a half-life marker and the reader sees how old it is, so the model itself judges how much to trust it. No scheduler is involved. This run does not test that half: neither benchmark varies time in a way that isolates it, and one thing at a time is the point of the rebaseline.

## 2. Where it came from

The archived version was the namesake of the whole programme. It had a seed alphabet of nine frames (stable fact, completed event, open concern, plan, and so on), six half-life markers from an hour to forever, and a rule that a working set of seven lines is shown to the model at a time. The notation was written by Haiku and read by Sonnet. The paper behind it claimed about eight tokens per memory against about fifty for a sentence, and “cross-model fidelity: confirmed”. Later generations let the model rewrite its own language and runtime, and that has not been carried forward.

The audit’s verdict was that this was the cleanest new idea in the archive and the least tested. The token claim was never measured: the code counted whitespace-separated words. The cross-model result was a few hand-picked questions with no score. The attempt on MemoryAgentBench truncated each context to 1,500 characters, so most of the facts a question needed were never written; one run scored zero rows, which is a failed run and not a zero; and a later generation scored 4 of 100 under a home-made judge through an adapter that skipped the memory altogether. None of it measured whether one model can read another’s notation, which is the claim that matters.

## 3. Hypothesis

Stated before running, in three parts, because they can fail separately.

1. Fidelity. Model B, reading a Glyph store written by model A with no primer, answers FactConsolidation questions as accurately as it does from the plain text of the same facts. The difference, paired, has a 95% interval that includes zero.
2. The file does the teaching. B reading A’s store scores about as well as A reading its own, and B reading the store with its header removed scores clearly worse. If it does not, the header is not what makes the language readable.
3. Tokens. Glyph stores each fact in fewer tokens than plain text, counted with a named tokenizer and with the header included.

The prior is that the third part is the weak one on this data. FactConsolidation facts are already short sentences, so there is not much to take out. The header costs about a hundred tokens once, which over the 455 facts of a 6K context is under a quarter of a token per fact, so even a small saving repays it. The question is whether there is a saving at all, and whether compression costs accuracy. The expectation is that it does, a little.

## 4. Design

**Store.** The seed alphabet as a header, then one line per fact. Model A writes each line from the plain fact. For LongMemEval a session becomes a block of lines and retrieval returns session ids, as for every other method.

**Reader.** Model B, from a different family, receives the official FactConsolidation prompt with the retrieved lines in place of the facts. The prompt does not mention Glyph, its syntax, or the header. There is no primer.

**Retrieval.** The Dense scorer runs over the Glyph lines themselves, at the same k as every method. That is deliberate. If compression removes what a retriever needs, the score should say so.

**Controls.** A plain-text store of the same facts. The same store read by the model that wrote it. The same store with its header removed, so B sees lines and no definitions.

**Round trip.** B is also asked to restate each retrieved line in plain English. The restatement is scored on whether the entities and the relation survive.

**Models.** Writer and reader are named in the run record and held fixed across all items. One pair first. A second pair only after the first has been reported.

## 5. How it is tested

The measurement is the cross-model one. The answer score is the official SubEM on B’s reply, cut to its first ten tokens, as for every FactConsolidation run. Two measures do not fit the shared columns and are reported as text beside the table: round-trip fidelity, and tokens per fact for Glyph and for plain text with the header amortised across the facts. Each is given with its interval and the tokenizer is named.

[MARCO?](/where-am-i) is the nearest existing evidence. Four model families, handed the same 45 public statements and the same encoder, placed six situations in the same place 34 times in 36, against 4 in 36 for shuffled labels. That test shared a primer and a vocabulary. Glyph is a stricter one: the only thing the two models share is the file.

MARCO?’s rule, that a belief must never be written as if it were evidence, is a natural extension for a memory language: a line could say whether the fact was checked or only recalled. The first run leaves it out so that the test measures one thing.

The standing rules for every Engram experiment apply here unchanged.

## 6. Results

## 7. What would change the conclusion

- If B scores the same with the header removed, the header teaches nothing. Either the notation is guessable or the questions never needed it, and “self-describing” has not been shown.
- If A reading its own store scores far above B reading it, what looked like a language was one model’s habit. The memory did not change hands.
- If plain text matches or beats Glyph on accuracy and on tokens, there is no case for the notation on this data, and the page will say so.
- If Glyph holds accuracy and saves tokens on FactConsolidation, the next step is to repeat it on LongMemEval and a second model pair before anyone should believe it.

## Sources

Experiment 10 of 10 in the Engram series. [← Canopy](/engram/canopy) · [Engram overview](/engram)

Glyph is one of ten experiments in Engram. Its rows come from one run on the pilot sets.