pi-fold 2.0.0 → 3.1.0

package/README.md

@@ -1,14 +1,12 @@
 # pi-fold
 
-[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.21856873.svg)](https://doi.org/10.5281/zenodo.21856873)
+[![npm](https://img.shields.io/npm/v/pi-fold)](https://www.npmjs.com/package/pi-fold) [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.22142456.svg)](https://doi.org/10.5281/zenodo.22142456)
 
-Agent-governed lossless context folding for Pi. A session folds its own transcript instead of losing it: the stale end collapses into short briefs while the exact originals stay on disk, addressed and verified by SHA-256, one tool call away.
+**Rotate stale text out of the context window without losing it, and spend the prefix cache once per commit instead of continuously.** The oldest part of a session collapses into short briefs while the exact originals stay on disk, one tool call away. Folding is automatic and needs no agent participation. The shipped shape is deterministic briefs paired with a durable memory store, which is the cheapest configuration measured that kept recall intact.
 
-The full write-up, with interactive versions of every mechanism on this page and the complete measurement story, is at [shaneconner.com/projects/pi-fold](https://shaneconner.com/projects/pi-fold/).
+![pi-fold folding a session in place: marks outline completed spans as the session works, and the marks commit together at a fold event](https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/fold-demo.gif)
 
-![pi-fold folding a session in place: the agent marks spans as it works, the marks commit together at a fold event, and a mark inside the protected fresh tail is held until it ages out](https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/fold-demo.gif)
-
-*The mechanism, drawn. One session runs as the transcript it is, user messages in cream, the agent's replies in green, tool results a quieter shade of the same green. The agent calls `pi_fold_context` with `action: "fold"` on spans it has finished with: each call records a mark that outlines the span it names and moves nothing, and the next fold event commits every standing mark in one rewrite, two outlined spans collapsing to two briefs at once while the ladder takes the unmarked stale end on its own. A dotted rule rides over the newest turns, the protected fresh tail: a mark that lands inside it is held at the fold event, so nothing folds out from under the turn still using it, and it commits a few turns later once it has aged out. The exact messages stay on disk, addressed and verified by SHA-256, and expanding a brief brings them back byte for byte. Illustrative animation, not a screen capture; the measured numbers are further down. A 6 MB mp4 of the same clip is at [media/fold-demo.mp4](https://github.com/shaneconner/pi-fold/blob/main/media/fold-demo.mp4).*
+*Marks outline spans as they complete and move nothing; the next commit applies every standing mark in one rewrite. Expanding a brief brings the exact messages back byte for byte. Illustrative animation drawn on the 2.x surface, where the agent laid the marks; since 3.0 the runtime lays them.*
 
 ## Install
 
@@ -16,235 +14,125 @@ The full write-up, with interactive versions of every mechanism on this page and
 pi install npm:pi-fold
 ```
 
-Node 22 or later, Pi 0.83 or later. Licensed under [MIT](./LICENSE).
-
-pi-fold serves the working tier of a four-tier memory stack. [pi-canon](https://github.com/shaneconner/pi-canon) is a separate, optional package serving the long-term tiers, the journal episodic and the canon semantic; the two compose but neither requires the other.
-
-## What it is
-
-pi-fold registers one tool, `pi_fold_context`, and gives the session's own agent nine actions over its transcript: `status`, `peek`, `fold`, `expand`, `refold`, `protect`, `unprotect`, `rebrief`, and `reboundary`, plus `unmark` under epoch scheduling. It also registers `/pi-fold-context` and `/fold-context` for the human.
-
-A fold takes a contiguous span of session entries and replaces it, in the window only, with a short brief. The entries themselves go to a fold store, byte for byte, addressed by their SHA-256 hash. The brief carries a handle, the handle resolves to the original, and expansion restores the exact bytes after verifying the hash. A fold is therefore a claim the runtime can check, not a claim the reader has to trust.
-
-Everything else follows from one invariant, and it is the reason several otherwise reasonable features are not here.
-
-**A provider prefix cache is positional.** It replays the longest byte-identical prefix of the prompt, so one changed byte at offset K discards everything cached after K. The cost of a mutation has nothing to do with its size: deleting a single character in the middle of a large window and rewriting a hundred thousand tokens near the end of it bill the same way, because both invalidate the same suffix.
-
-**So the runtime may move a byte in the window only at a moment it is already rewriting the window**, which is a commit or a fold. Everything else is a pure append at the tail, and an appended byte is never later altered, shortened, removed, or repositioned. Stated plainly: never mutate the window, and never show something and then take it back. The second half is the one that costs features. Anything that displays state and then updates it in place is out, however useful the display was, because updating in place is a mid-prefix write with better manners.
-
-**Marks are free, because a mark moves nothing.** A `fold` action records a pending mark and returns immediately. The mark lives in durable session state, outside the window; the projection stays byte-identical and the cached prefix survives. The agent can therefore mark as it works, the moment a span stops mattering, without weighing each fold against a cache invalidation.
-
-**A commit is the one moment bytes move, and every accumulated decision lands at once.** It goes through the preparation and commit machinery every fold has always used, so a committed mark produces the ordinary fold record. Committing is the runtime's call, not a verb the agent can spend: measured 2026-08-07, an agent given the verb called it twice and the runtime correctly held it both times, which is surface without function.
-
-**Folds are lossless, and there are two ways back in.** Folds form a DAG whose leaves are the exact session entries, and every fold record keeps ordered references down to those leaves. `peek` is the cheap read: it returns one fold's exact SHA-256-verified source as a tool result, at any depth, with the ancestors still collapsed, without touching the projection. It takes `offset` and `bytes` for a bounded slice, and any child fold id is peekable, so a large fold has a narrow read. `expand` is the commitment, restoring that source in place until the fold is refolded, outside-in and one level at a time. Looking is not the same as taking, and most of the time an agent needs to check one detail rather than reopen an hour of work.
-
-When the agent does not act, an automatic ladder manages the window the same lossless way, and it is deliberately more conservative: completed tool batches, structurally closed chapters, and consolidation of existing folds only past explicit thresholds. Only the oldest eligible context changes; the fresh window is never touched. A session that never calls the tool degrades gracefully into lossless hierarchical compaction.
-
-## Folding against compaction
-
-Compaction does not continue a session, it starts a new one. Pi's native compaction summarizes the transcript, discards the originals, and begins again from the summary. In the run measured below that is literal rather than rhetorical: native compacted three times, and the request immediately after each one carried a prompt of exactly zero tokens, 369.0k to 0k, then 364.6k to 0k, then 363.0k to 0k. A continuation does not restart at zero. The handoff is also written before the end, so the freshest material is exactly what is missing from it, and what a summarizer drops as bookkeeping is what becomes unrecoverable.
-
-Folding keeps the turn. Occupancy crosses a threshold, spans fold in place, and the session continues holding a verified copy of everything it folded. The window stays inside a band instead of sweeping the whole allotment: the same measured run put native between 0k and 369.0k across 117 requests, with 12 of them served above 300k, while the folding arm ran between 1.1k and 236.9k across 82 requests and never entered that zone. Roughly the same mean window, 140.9k against 138.8k; what differs is the range. That matters because behavior is not constant across it. You are not running one system at two sizes; you are running a different system at 40k than at 400k.
-
-## Measured results
-
-The first campaign behind these tables contains two experiments. The first, here, is pi-fold against Pi's native compaction. The second, further down under [epoch fold scheduling](#epoch-fold-scheduling), is pi-fold against itself with fold scheduling as the only variable. A later transcript-only hidden-mass campaign is reported separately below, with its own protocol and claim limits.
-
-One 64-stage staged assignment over the curl C repository: one Pi session per run, one user message, and the agent calls a `repo_stage` tool 64 times inside a single agentic turn. Model gpt-5.6-sol at xhigh effort, provider openai-codex. The 272,000-token figure in these tables is the transport's per-request input descriptor, which is what pi-fold budgeted against; the Codex catalog actually serves gpt-5.6 at a 372,000-token cap, which is where the native arm lived, peaking at 369.0k before each compaction. One arm runs pi-fold, the other runs Pi's native compaction.
+Node 22 or later, Pi 0.83 or later. MIT licensed.
 
-There is one run per arm, and native did not finish: it ended its turn at stage 56 of 64. That leaves two pairings, and both are reported here rather than the flattering one.
+Folding does not require a memory store, but every best result measured came from pairing it with one. The campaigns used [pi-canon](https://github.com/shaneconner/pi-canon), installed separately. The two began as a single working setup and were split into separate extensions for modularity, and the experiments are what established that they work better as complements than either does alone. Only that pairing has been measured. Any durable long-term memory store is expected to compose the same way, though that is a suspicion rather than a result.
 
-The measurement is written up as a paper, with the figure sources, the redacted per-request ledgers, and the campaign log deposited beside it: [doi.org/10.5281/zenodo.21856873](https://doi.org/10.5281/zenodo.21856873). The story of how the campaign got here is on [Medium](https://medium.com/@shane.conner/compaction-doesnt-have-to-mean-starting-over-89c0b319d1a6).
+## What it does
 
-### Same seed, both cut at stage 56
+Compaction replaces the transcript with a summary and discards the originals. Folding replaces stale spans with briefs and keeps the originals retrievable.
 
-**Basis:** each arm is cut at the ledger record that delivered the stage 56 payload, and usage and cost are summed from the same billed records at or before that cut, including native's three compaction records, which all precede it. Same seed, same plan hash `6ed7c5f9e957`, both runs on 2026-08-07.
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/fold-vs-compaction-dark.svg">
+  <img alt="Compaction turns the transcript into a summary and discards the originals, so the session starts over. pi-fold turns it into a brief and keeps the exact originals on disk, one peek or expand away, so the session keeps going." src="https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/fold-vs-compaction-light.svg">
+</picture>
 
-| Measure | pi-fold | native | ratio |
-| --- | ---: | ---: | :--- |
-| Total tokens | 9,598,126 | 16,011,169 | 0.60x |
-| Fresh input | 1,249,415 | 1,322,539 | 0.95x |
-| Cache read | 8,309,760 | 14,602,240 | 0.57x |
-| Output | 38,951 | 86,390 | 0.45x |
-| Billed calls | 69 | 114 | 0.61x |
-| Cost, Pi's own ledger | $11.57 | $21.00 | 0.55x, a 44.9% saving |
-| Pooled cache share | 0.869 | 0.917 | inverted, see below |
-| Peak request context | 236,861 | 369,024 | threshold 272,000 |
-| Long-context surcharge | $0.00 | $4.50 on 17 requests | 21.4% of native's bill |
-| Compactions | 0 | 3 | |
-| Agent file reads | 27, 0.06 MB | 117, 0.93 MB | 4.3x the reads, 15x the bytes |
-| Wall clock | 25.2 min | 42.7 min | 0.59x |
+A fold takes a contiguous span of session entries and replaces it, in the window only, with a short brief. The entries go to a fold store byte for byte, addressed by their SHA-256 hash. The brief carries a handle, the handle resolves to the original, and expansion restores the exact bytes after verifying the hash.
 
-### Both complete, 64 of 64 stages
+The package registers one tool, `pi_fold_context`, with nine actions over the transcript: `status`, `peek`, `brief`, `expand`, `refold`, `pin`, `unpin`, `reboundary` and `unmark`. A session that never calls the tool still folds; it degrades into lossless hierarchical compaction.
 
-**Basis:** whole-run basis for both arms. Usage and cost are summed from every billed record in the full session ledger, message records plus native's three compaction records; the ledger is the source of truth for both tokens and dollars. The two runs sit on different days, so provider conditions differ.
+## Results
 
-| Measure | pi-fold | native | ratio |
-| --- | ---: | ---: | :--- |
-| Total tokens | 11,435,313 | 16,970,754 | 0.67x |
-| Fresh input | 1,489,599 | 1,407,978 | 1.06x |
-| Cache read | 9,895,936 | 15,510,016 | 0.64x |
-| Output | 49,778 | 52,760 | 0.94x |
-| Billed calls | 82 | 108 | 0.76x |
-| Cost, Pi's own ledger | $13.89 | $21.18 | 0.66x, a 34.4% saving |
-| Pooled cache share | 0.869 | 0.917 | inverted, see below |
-| Peak request context | 236,861 | 368,639 | threshold 272,000 |
-| Long-context surcharge | $0.00 | $4.81 on 20 requests | |
-| Fold events / compactions | 6 fold events, 0 compactions | 3 compactions | |
-| Agent file reads | 39, 3.84 MB | 108, 4.62 MB | 2.8x the reads |
-| Wall clock | 34.2 min | 32.4 min | 1.06x, pi-fold slower |
+One 64-stage assignment over the curl C repository, run as a single agentic turn. Same model, same 251,520-token serving budget, every run. After all 64 stages landed, one withheld message asked sixteen questions about material from across the whole session.
 
-**Where the cost numbers come from.** Cost is Pi's own per-request ledger accounting, not a rate card chosen for the writeup. Every billed record's cost equals that record's own usage priced at the tier Pi applied: base rates of $5.00/M input, $0.50/M cache read and $30.00/M output, and a long-context tier of $10.00/M, $1.00/M and $45.00/M on any request whose context exceeds the model's 272,000-token window. That reproduces the ledger to the cent on both arms, including pi-fold's complete-run total of $13.89. Reasoning tokens sit inside output and are not billed separately. Native's three compaction summarizations are themselves billed calls, $0.58 at stage 56 and $0.54 on the complete run.
+| Condition | Correct of 16 | Wrong | Cost | Fresh input | Cached input | Wall clock |
+| --- | ---: | ---: | ---: | ---: | ---: | ---: |
+| Compaction | 4 | 3 | $151.24 | 7.64M | 160.2M | 6.9h |
+| Compaction + memory store | 14 | 0 | $201.70 | 9.65M | 231.7M | 9.0h |
+| pi-fold, deterministic briefs | 11.3 avg | 3.5 avg | $93.65 | 5.74M | 91.2M | 4.6h |
+| **pi-fold + memory store** | **14** | **0** | **$100.85** | **5.57M** | **102.2M** | **5.2h** |
+| pi-fold, agent-written briefs | 3 | 9 | $111.96 | 7.12M | 108.8M | 5.4h |
 
-**Caveats, because the numbers rest on them.**
+*Fresh input is what the provider had to read for the first time; cached input is what it replayed. The deterministic-briefs row averages four runs that scored 14, 14, 9 and 8.*
 
-- **Fresh input is essentially identical.** 0.95x at stage 56 and 1.06x on the complete pairing: neither runtime consistently reads less new material. The 6.4 million token gap at stage 56 is re-sent accumulated context, 6.29 million of it cache read alone, not a difference in how much material was consumed.
-- **No wall-clock claim on the complete pairing.** pi-fold was 5.6% slower there, 34.2 minutes against 32.4. The wall-clock win belongs to the stage-56 pairing only.
-- **The cache share runs backwards, and that is the useful finding.** Native holds the higher pooled share, 0.917 against 0.869, while spending 67% more tokens and billing 82% more dollars, because cache share measures how cheap each prompt was rather than how much prompt was needed.
-- **Part of the cost gap is a pricing tier, not token volume.** At flat base rates native's stage-56 bill would be $16.51, a 29.9% saving rather than 44.9%, and its complete-run bill $16.38, a 15.2% saving. The tier is real money a native run really pays, but quote the flat-rate figures when the argument is about tokens rather than price schedules.
-- **One run per arm.** Native never finished; the supervisor lost the worker at the stage-57 request, so the stage-56 cut excludes native's final 6 messages and $0.57 of ledger cost. The complete pairing ran on different days. The stage-56 pairing is same-day and is the stronger claim.
-- **The agent never called the tool.** All 62 folds in that run were the automatic ladder's; every commit records zero agent marks. So these are the fallback's numbers, produced by the part of the design that is closest to prior work, with agent curation, the distinguishing claim, unexercised.
-- **The reread story is a count, not a duplicate-payload metric.** It rests on file read count and bytes from disk, both in the tables above.
+Compare the two store-paired rows. Both scored 14 correct with none wrong, and folding cost half. That gap is request count rather than price: compaction needed roughly twice the provider requests for the same work, plus summarizer calls that never show up in its own ledger.
 
-Full method, artifact hashes and per-iteration history: [docs/fold_vs_compaction/experiment-log.md](https://github.com/shaneconner/pi-fold/blob/main/docs/fold_vs_compaction/experiment-log.md). Queued work: [docs/next-steps.md](https://github.com/shaneconner/pi-fold/blob/main/docs/next-steps.md).
+Compaction alone abstained nine times and was wrong three, so its dominant failure was declining rather than inventing, but it did invent. And the fold's deterministic runs are bimodal rather than average: two answered 14 with nothing wrong, two lost old material, and cost did not predict which. The store was tried once on each arm and both came back with nothing wrong, which is suggestive, not a rate.
 
-### Transcript-only hidden mass, protocol v4
+Agent-written briefs scored worst of anything measured, which is why that invitation ships off.
 
-The follow-up campaign plants seeded values only in the transcript, never in the repository checkout, and withholds one final 30-cell block until all 64 stages are delivered. The plan, question order and target checkout were frozen before the readout. Both arms ran gpt-5.6-sol at xhigh effort on the same plan hash, `eb488827c46d`.
+Four papers with figure sources, redacted per-request ledgers and campaign logs are on Zenodo: [design and trace evaluation](https://doi.org/10.5281/zenodo.21856873), [working memory under context shedding](https://doi.org/10.5281/zenodo.21980746), [ephemeral retrieval](https://doi.org/10.5281/zenodo.22142454), and [rotation and retention](https://doi.org/10.5281/zenodo.22142456), which is the campaign above. Narrative versions on [Medium](https://medium.com/@shane.conner/a-compaction-summary-is-one-record-doing-a-stores-job-3f46212ef059), interactive versions at [shaneconner.com](https://shaneconner.com/projects/pi-fold/), and the full chronology in [the experiment log](https://github.com/shaneconner/pi-fold/blob/main/docs/fold_vs_compaction/experiment-log.md).
 
-| Outcome across assigned attempts | pi-fold | native compaction |
-| --- | ---: | ---: |
-| Completed 64 stages | 2 of 2 | 0 of 2 |
-| Join endpoints recorded correctly | 8 of 8 | 0 of 8; all eight recorded `unknown` |
-| Ordinary probes | 40 of 42 | not available from two incomplete runs |
-| Withheld end-block cells | 60 of 60 | not reached |
+## Limits
 
-Native compaction lost a different continuation key in each attempt, after stages 39 and 56. The first run was killed after an unbounded resume defect produced 1,761 provider responses; the second carried the fixed three-resume bound and failed by name within 33 minutes. At that point the campaign stopped native rather than launching until one survived, which would select on the outcome, and ran a second pi-fold attempt to leave two attempted assignments per arm.
-
-Both pi-fold end-block turns included context recovery. The agent made five context calls in repetition 1 and nine in repetition 2 before answering. At the answering requests, the corrected carriage sweep classified 53 of 60 expected end-block values as visible raw and seven as recoverable but not visible, all seven in repetition 1. Those seven answers were correct, but the final projection does not identify how their bytes reached the answer: 60 of 60 is the scored outcome, not proof that recovery made the whole ledger visible. All eight withdrawn correction values were visible raw, and the agent selected the correction every time. The same 102-row certification found one ordinary miss with its expected value visible in a brief and the other recoverable but not visible, while the same matched control in both runs had its value absent from the answering branch.
-
-There is no direct cross-arm end-block score and no cost ratio from this campaign: neither native run completed, and native repetition 1 is excluded from cost comparison because of the harness runaway. Two attempts do not estimate a population failure rate. The portable result contains all 102 rows and the source artifact hashes: [hidden-mass-results.json](https://github.com/shaneconner/pi-fold/blob/main/docs/fold_vs_compaction/hidden-mass-results.json). The full chronology and stop rule are in the [experiment log](https://github.com/shaneconner/pi-fold/blob/main/docs/fold_vs_compaction/experiment-log.md).
+- **Native compaction is strong and is the right default for most sessions.** It needs no installation and no tool surface. This is an alternative for workflows where the numbers above matter, not a replacement.
+- **The benchmark is narrow by construction.** It was built to reproduce specific things compaction handles badly, one to four runs per condition, and it says nothing about production behavior.
+- **Folding is not a memory store.** It rotates transcript source out of the window; durable placement of what a session learned is a separate job, and the pairing is where the measured value sat.
 
 ## How it works
 
-pi-fold maps canonical session entries into a fold lattice. Collapsed folds double as a browsable index of earlier work: the briefs sit in the stable prefix of the window, which providers typically serve from cache, so the agent can page through them at little cost, note which spans matter to its current task, expand those, and leave the rest folded. `status` with `detail: "tree"` lists every fold, nested ones included, in transcript order with its depth and parent.
-
-The agent's own actions come first. Through the context tool it folds any structurally closed span it judges stale, folds completed tool batches, consolidates two or more adjacent folds into a deeper one, expands any fold whose detail is needed again, re-briefs a fold whose summary reads wrong, dissolves a mis-cut boundary with `reboundary`, and protects entries that must stay raw. None of the ladder's pressure or width thresholds apply to these actions. Curation the agent cannot fix afterwards is curation it will not risk making, which is why the two correction verbs are part of the ordinary surface rather than an extra.
-
-Every fold, agent-made or automatic, obeys the same structural rules. Folds nest rather than partially overlap. Chapter folds align to closed user, assistant and tool-batch units, consolidations take two or more adjacent folds, and protected or fresh evidence never folds.
-
-The automatic law folds only completed tool batches and structurally closed chapters, absorbs only the tool-result folds inside a chapter's own span, and treats fold placeholders as ordinary span material once the window carries `consolidateAfter` unpinned folds, which keeps the brief index tidy long before pressure matters. Briefs are bounded at 2,000 characters each, one cap for supplied and runtime-written briefs alike, so the visible index normally costs a low single-digit share of the window. Automatic foldability is membership, not position: a span folds because it is a completed tool batch or older material a chapter can compose, and what holds it back is the agent's pins, the guaranteed-raw fresh tail, the blacklist, and the commit-time guard on the open turn.
+### The cache invariant
 
-Below `maxTarget` occupancy of the serving budget, 80% by default, the runtime is quiet: nothing folds automatically at all, because a commit is an epoch transition that risks the whole prefix cache, and the cadence has to be the fewest commits that still keep the window inside its budget. The remaining fifth of the budget is the runway the commit itself spends. Every occupancy number in this document is a share of that one serving budget: there is a single denominator, `capacityAccounting.budgetTokens`, and no threshold is stated against the raw provider window.
+**A provider prefix cache is positional.** It replays the longest byte-identical prefix of the prompt, so one changed byte at offset K discards everything cached after K. The cost of a mutation has nothing to do with its size.
 
-Every brief is deterministic, written by the runtime from the span itself with no model call: a tool fold carries the call, its arguments, the head and tail of the result, and the agent's own notes from the batch; a chapter carries its opening text and every assistant note in span order; a consolidation indexes each child. A model-written brief generator shipped through 1.0.x and was deleted on its own measurement: across 15 sealed runs it wrote 1,186 upgraded briefs, 75 percent of which were never consulted by the agent or never became visible before the session ended, and a build with the generator removed scored the same recall for less money. The agent that wants better words on a fold writes them itself: `fold` accepts a brief, `rebrief` rewrites one in place, and a supplied brief is kept verbatim. Sessions already holding model-written briefs keep them on load.
+**So the runtime may move a byte in the window only at a moment it is already rewriting the window**, which is a commit or a fold. Everything else is a pure append, and an appended byte is never later altered, shortened, removed or repositioned. Never mutate the window, and never show something and then take it back. The second half is the one that costs features: anything that displays state and then updates it in place is out, because updating in place is a mid-prefix write with better manners.
 
-The serving budget is what the deployment states it may fill, because the per-request max-input descriptor a provider advertises assumes a full output reservation and understates the real ceiling. With no `providerInputBudget` supplied the runtime falls back to the transport's descriptor and estimates a reservation out of it, which is the conservative default for a model it has no budget fact for; the descriptor is reported in status either way, so the gap stays auditable. A provider-window fence aborts an unsafe request before transmission if no lossless fold can make room, and says so out of band rather than by writing into the window.
+### Marks and commits
 
-Pi's automatic compaction stays enabled underneath pi-fold; there is no setting to turn off. The runtime intercepts every compaction pass and answers it by kind. A threshold pass is cancelled outright, because the fold commit at `maxTarget` fires first and folding remains authoritative. An overflow pass, the one Pi starts when the provider rejects a request as too large and is prepared to retry, becomes a recovery in the shape of `/tree`, automated: the failed path is labeled with its lineage, the session branches back to the last message before the error, and the retried pass folds and commits before it transmits, so the re-issued message fits inside a window that kept its folds and its exact originals. The agent gets one notice that a rollback and commit happened. A trailing assistant message still holding unanswered tool calls is not replayed, and the notice says so. Manual `/compact` passes through untouched as the user's escape hatch. Turning Pi's automatic compaction off removes the hook this lane arms on: the runtime still classifies an overflow when it appears, but reports that it cannot roll back rather than guessing at a tree whose agent state disagrees with it.
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/marks-and-commits-dark.svg">
+  <img alt="A completed unit becomes a pending mark that leaves the window and its cached prefix untouched. Marks repeat for free below maxTarget; at maxTarget one commit applies every standing mark in a single rewrite. From the brief, peek appends the exact bytes and expand restores them in place, both after verifying the SHA-256." src="https://raw.githubusercontent.com/shaneconner/pi-fold/main/media/marks-and-commits-light.svg">
+</picture>
 
-Evidence ingestion writes read-only artifact files under the session directory's `pi-fold-evidence/`, sets them to mode `0444`, and enforces a 512 MB session cap. These immutable files provide exact-recovery anchors for oversized tool results. It is always on and there is no setting to turn it off: the artifacts are what an oversized result folds against, so a deployment that stopped writing them would keep its folds and lose what makes them lossless.
+A mark moves nothing. The moment a tool batch or chapter closes, the runtime cuts it into a pending mark, up to eight cuts per pass, stalest first. The mark lives in durable state outside the window, so the projection stays byte-identical and the cached prefix survives. Marks accumulate for free until one commit applies all of them in a single rewrite.
 
-## Epoch fold scheduling
+What the agent owns is the words. `brief` writes its own sentence onto a pending fold for free, because that brief reaches the window only when the commit writes the placeholder for the first time. A supplied brief is kept verbatim; an unbriefed fold commits with the deterministic brief the runtime writes from the span itself, and the provenance records which happened. Curation stays correctable until the commit: `reboundary` re-cuts a span, `unmark` withdraws a pending mark, and a pending brief can be rewritten any number of times at no cost. A standing fold's brief is never rewritten, because that rewrites the projection at its placeholder.
 
-Folding is cheap to decide and expensive to apply, and the expense has nothing to do with how much a fold saves. The bill is set by how often the projection changes, not by how many tokens each change reclaims. A session that folds once per turn re-sends its prefix once per turn; a session that folds three times in total re-sends it three times.
+### The band
 
-Epoch scheduling is the only scheduler, and it splits folding into the two phases described above: a `fold` action records a free pending mark, `unmark` withdraws a standing decision, and a commit applies every pending mark in one projection rewrite. Applying each fold at the moment it was made shipped as an option through 1.0.2 and was deleted once measured: on the pairing below it cost 54 prefix rewrites where epoch paid 3.
+A commit fires on one condition: occupancy reaching `maxTarget` of the serving budget. It applies marks stalest first and stops at `minTarget`, so the newest events stay raw and a mark past the bound waits for the next epoch. Below `maxTarget` the window is quiet, because a commit is an epoch transition that risks the whole prefix cache and the right cadence is the fewest commits that keep the window inside its budget.
 
-Two runs from the campaign put a number on the difference. This is the second of the two experiments: pi-fold against itself, both runs on gpt-5.6-luna, same plan, same seed, same configuration with fold scheduling as the only variable. It is internally controlled, and its numbers are not comparable across to the gpt-5.6-sol pairings above. Applying folds immediately spent 19,623,502 fresh input tokens across 105 requests at a pooled cache share of 0.119. Recording marks and landing them together spent 3,603,440 across 99 requests at 0.756: a 5.4x reduction in the tokens the provider had to read for the first time, carrying the total from 22,316,589 tokens down to 14,818,833. Epoch was slower in wall clock on that pairing, 33.5 minutes against 26.7, so the trade bought tokens and not time, and it is one run per arm rather than a rate.
-
-The agent is meant to lead here: mark as you work, and let the runtime pick the seam. The automatic law is the fallback, exactly as it is for folding itself. Its own decisions become marks instead of folds, and it commits the epoch on one condition: occupancy reaching `maxTarget` of the serving budget. Nothing else fires a commit automatically. Each automatic commit carries a target: it folds down toward `minTarget`, so if the agent's marks would not reach that line the law tops the epoch up with the stalest unprotected spans until they do, and the commit result records which marks were the agent's and which were the top-up. The gap between the two lines is the hysteresis, which is what makes the spacing between events structural rather than hoped for.
-
-Three refinements keep the two phases from getting in the way. A fold whose span sits within the last few messages of the window applies immediately even in epoch mode, because almost nothing follows it to invalidate. An `expand` issued while marks are pending opens the commit epoch, so a restore plus a batch of folds costs one rewrite between them rather than two; `peek` stays immediate, since it appends a tool result and never edits the prefix. And a peek result is itself a completed read, so at the next commit it folds away automatically unless the agent committed to what it saw, which keeps look, decide, discard nearly free.
-
-Protection is unchanged. Top-up never marks protected evidence, and a pending mark that has become protected refuses at commit with a message naming it, leaving the rest of the epoch to apply. `pi_fold_context {"action":"status"}` reports the pending marks under `automatic.scheduling` with their origin, the estimated share of the serving budget a commit would free, and the estimated prefix tokens it would rewrite.
+```
+272,000  the transport's descriptor
+─────────────────────────────────────────────────────
+~250k    peak: inflow continues while the epoch computes
+217,600  maxTarget 0.80   ← the commit fires here
+   ⋮                        marks accumulate, free, quiet
+ 54,400  minTarget 0.20   ← the commit stops here
+─────────────────────────────────────────────────────
+```
 
-## Reclaiming peek output
+Every commit re-reads roughly `minTarget` of the budget uncached, so a shallower cut buys less runway for nearly the same bill: the rewrite tax is `minTarget / (maxTarget - minTarget)`, 0.33 at 0.80/0.20 against 1.0 at 0.80/0.40.
 
-A `peek` returns a fold's exact stored source, bounded by the chapter cap. That copy then sits in the window as raw evidence while the fold it came from sits beside it as a placeholder: the same bytes, held twice, and the copy is the one the ladder cannot take. A runtime that does not classify a peek batch as a completed read measures the pressure at every rung and finds nothing eligible. On a workload that peeks freely this is the whole failure: a measured 64-stage session ended with 14 raw peek results holding 1.9M characters, 82% of everything still unfolded, while the ladder had already folded 53 of its 56 stage results and had no supply left.
+### Peek and expand
 
-So a peek result is classified as a completed read, unconditionally. The peek batch is eligible for the same tool rung under the same protections: the fresh tail still stays raw, protected evidence is still refused, and the fold record is the ordinary reversible one, so the source recovers exactly. Nothing about the ladder's thresholds changes; only what counts as foldable does. The cost is that a peeked span the agent is about to work from can fold back before it is used, which means peeking again to read it; the duplicate window it otherwise leaves behind is the larger bill.
+`peek` returns a fold's exact SHA-256-verified source as a tool result at any depth, with the ancestors still collapsed. It appends and never edits the prefix, so checking what a fold holds costs nothing a commit has to pay for. It takes `offset` and `bytes` for a bounded slice and reads any child fold by id, so a large fold has a narrow read.
 
-## What the invariant cost
+A peek is ephemeral by default: the bytes ride the window until the model's next message, then their place holds a one-line placeholder and the reply that used them is the surviving trace. Peeking again is lossless and costs one append. The default follows what the agent chose: on the run measured for it, six of seven peeks asked for ephemeral at first exposure. `expand` is the commitment, restoring the source in place until the fold is refolded. Looking is not the same as taking.
 
-Three separate features shipped that each rewrote the window mid-prefix, and each carried a code comment asserting the edit was tail-local and therefore safe. In all three cases the comment was the marker of the bug, and it was also the reason the bug survived review: the reasoning looked done, so nobody redid it.
+### Clipping, pins and the compaction path
 
-Peek reclamation was the costliest. It removed duplicate bytes from the window after a peek and deferred the removal until a later assistant message existed, so the edit would land behind the tail. That deferral is what makes it look tail-local and what guarantees it is not: waiting for a later message is precisely the condition under which the window has already grown over the site of the edit. Measured 2026-08-07, two such reclamations cost 100k fresh tokens. A peek is now append-only, and its duplicate bytes are reclaimed by the tool-fold rung at the next commit, the one moment a rewrite is already being paid for.
+Tool results get a second, lighter treatment at the same commit. Results in the oldest `toolFoldThreshold` share of the projected window, half of it by default, are clipped in view: the identifying head stays, the rest becomes a marker naming the entry id that peeks it back, and the stored bytes are untouched, so a later fold of that span still encodes the full result.
 
-The ephemeral surfacing slate went the same way, and it deserves its own note because the idea was not the problem. It scored the fold set against the live window and offered the agent a short slate of folds worth peeking at or expanding, and the selection was useful. What failed was the delivery: the slate was shown on one pass and withdrawn on the next so a recomputed slate could take its place, which is a show-then-retract. Its bytes occupy prefix positions in one request and different bytes occupy them in the next, so the cached prefix dies. What is deleted is the delivery, not the selection: the per-request carrier is gone and so is the external suggestion-source channel that fed it, which was a registration into a carrier that no longer exists. The selector survives, and the delivery that replaced it rides the commit boundary: at most one suggestion, carried as literal persisted bytes where the rewrite is already being paid for. A fold is suggested only when its content matches the current work while its brief does not say so, which points the agent at exactly the fold whose placeholder undersells it; a fold whose suggestions are ignored twice with no takes is silenced permanently.
+Briefs are bounded at 2,000 characters each, so the visible index normally costs a low single-digit share of the window. Folds nest rather than partially overlap, and collapsed folds double as a browsable index: `status` with `detail: "tree"` lists every fold in transcript order with its depth and parent. `pin` holds entries raw through every fold and frees nothing, so pinned mass makes the rest of the window fold sooner; a pinned-share cap stops that from becoming a window nothing can reclaim.
 
-Anticipatory guidance went with them: pressure milestones, the live advisory, curation reminders, the last-call notice. Guidance about a coming event has to arrive before the event, so it can never ride a rewrite the runtime was already paying for; it has to create one. Eleven runs also say the warning bought nothing: voluntary fold share was 0.00, and the three runs built specifically to invite curation produced 0, 1 and 0 voluntary folds. What is left in the window is a bounded receipt of what the runtime just did, built only when a commit produced one, landing inside the freeze, never rendered twice.
+Pi's automatic compaction stays enabled underneath and is intercepted by kind. A threshold pass is cancelled, because the fold commit fires first. An overflow pass becomes an automated recovery: the session branches back to the last message before the error, and the retried pass folds and commits before transmitting. Manual `/compact` passes through untouched as the escape hatch.
 
-Earlier releases shipped truthful capacity accounting, admission control on `peek` and `expand`, retained pending marks, stage-identified fold briefs, the current-turn commit guard, the pinned-mass backstop, the status index diet and the append-only projection itself as individually opt-in levers. All of them are now simply how pi-fold works, and the flags are gone rather than deprecated; git history is the lineage.
+Evidence ingestion writes read-only artifacts under the session directory's `pi-fold-evidence/` at mode `0444`, under a 512 MB session cap. It is always on: the artifacts are what an oversized result folds against, so a deployment that stopped writing them would keep its folds and lose what makes them lossless.
 
-`pi_fold_context {"action":"status"}` reports real projection rewrites separately from observed provider-side cache misses, per-message digests included: a miss on a projection the runtime did not rewrite is provider-side by construction, and conflating the two hides exactly the bug this invariant exists to prevent. On the complete 64-stage run that instrument read 76 pure appends, 6 rewrites, and 0 unattributed rewrites, with the six rewrites landing batches of 12, 14, 9, 11, 9 and 7 folds.
+## Configuration
 
-## Relation to prior work
+`registerPiFold(pi)` works with no arguments, and the defaults are the configuration the campaign above actually ran: deterministic briefs, the invitation off, the tool-call diet at half the window, and a 0.80/0.20 band. Nothing here needs tuning.
 
-pi-fold combines two recent lines of work in one mechanism. Lossless hierarchical compaction, exemplified by LCM ([arXiv:2605.04050](https://arxiv.org/abs/2605.04050)) with an interactive walkthrough at [losslesscontext.ai](https://www.losslesscontext.ai/), builds a summary DAG over older messages with lossless pointers to every original, but the system alone decides when to compact and the folds accrete from the beginning of the conversation as it ages, so the cuts track position. Self-GC ([arXiv:2607.00692](https://arxiv.org/abs/2607.00692)) treats context as indexed, recoverable objects with fold, mask and prune actions, proposed by a side-channel planner under harness enforcement. In pi-fold the session agent itself holds the verbs, in-band, so the cuts can track relevance instead; the automatic ladder is the fallback, and a session that never touches the tool degrades into lossless hierarchical compaction.
+The one thing worth adding is a long-term memory store. Rotation and retention are different jobs: folding gets stale material out of the window, a store keeps what the session learned. A handoff summary tries to do both in one record, written once under time pressure, bounded by the window it has to fit in. Folding is the rotation half and only that half.
 
-Context-Folding ([arXiv:2510.11967](https://arxiv.org/abs/2510.11967), ByteDance Seed) has the agent branch into a sub-trajectory and fold it on completion, with the behavior shaped by reinforcement learning; the fold keeps a summary of the outcome rather than a lossless path back. MemGPT and its successor [Letta](https://github.com/letta-ai/letta) are the canonical agent-held memory-verb systems: the agent edits its own memory blocks and archives while the window pages against external stores. Two things separate pi-fold from that line. First, what is governed: Letta manages traffic between the window and outside storage, and when its message queue overflows the in-window record becomes a lossy recursive summary; pi-fold restructures the transcript in place. Second, recovery: theirs is retrieval, a search that re-ingests whatever comes back as new tokens; ours is expansion, which restores the exact folded entries after SHA-256 verification.
+Six options, and every other name is refused.
 
-The placement is narrow. The lossless DAG is shared ground. The distinguishing claim is agent curation, and the measured run never exercised it, so the numbers above belong to the fallback. The part that is new is stated here and not yet measured.
+| Option | Default | Effect |
+| --- | --- | --- |
+| `thresholds` | `{ maxTarget: 0.80, minTarget: 0.20, consolidateAfter: 10, minFoldChars: 8000 }` | The band, set whole or not at all. The first two are proportions of the serving budget. `consolidateAfter` is how many visible roots a parent is owed. `minFoldChars` is the size a fold has to reach to be worth making, and the floor a gap between folds has to clear to stay raw. Validated atomically; an impossible policy is refused by name rather than clamped. |
+| `providerInputBudget` | the transport's descriptor | The tokens this deployment may actually put in a request, already net of any output reservation. Every ratio, fence and budget on this page is a share of this one number. |
+| `blacklistAutoFoldTools` | `new Set()` | The exception list. Every completed tool batch is foldable without an agent mark; name a tool here and its results stay raw. |
+| `toolFoldThreshold` | `0.50` | The tool-call diet: a share in `[0, 1)` naming the oldest fraction of the window whose tool results are clipped in view. `0` turns it off. |
+| `postFoldNotice` | `false` | The invitation switch. False is the shape that won the campaign above: no carrier invites a brief and every fold ships the runtime's own words. `true` appends a standing invitation to improve a pending brief. |
+| `workingMemory` | off | An in-window keyed dictionary the agent maintains beside the fold index, with `remember` and `recall` added to the tool. Offered to one run, which never called it. |
 
-The design also pairs with Pi's session trees: a fold's collapsed-or-expanded state is a natural unit for curating exactly which context a subagent or workflow leg inherits.
+Four commands ship for the human: `/fold-status`, `/fold` (commit every staged mark now), `/fold-editor` (the interactive map of the window) and `/fold-settings`.
 
-## Configuration
+Upgrading from 2.x: the `fold` action is gone from the tool surface, since the runtime stages every completed unit itself; `guidance` is refused by name and `postFoldNotice` replaces it; `thresholds.freshTail` is gone, and a settings file carrying it is migrated with every tuned value kept. Existing sessions stay readable, folds and briefs included.
 
-The package entry calls `registerPiFold(pi)` with the defaults below, and that is the whole surface: four options, and every other name is refused with the replacement named in the message. A host still passing `summarizer` is refused with the deletion message rather than silently ignored. Hosts that call the named `registerPiFold` export directly may pass an options object. Replacement sets replace the defaults; they are not additive.
+## Prior work
 
-| Option | Default | Effect |
-| --- | --- | --- |
-| `thresholds` | `{ maxTarget: 0.80, minTarget: 0.20, freshTail: 0.02, consolidateAfter: 10 }` | The thermostat, set whole or not at all. The first three are proportions of the serving budget; `consolidateAfter` is a positive integer count. Validated atomically at registration, and an impossible policy is refused by name rather than clamped. User policy only: no agent action can read it back as a mutable surface, though status reports the values. |
-| `providerInputBudget` | the transport's descriptor | The tokens this deployment may actually put into a request, already net of whatever output reservation it holds back. Every ratio, fence and budget divides by it directly: pi-fold subtracts nothing from a declared value. Supply it when the deployment knows its real ceiling, because the per-request max-input descriptor a provider advertises assumes a full reservation and understates it. With no value supplied the runtime falls back to that descriptor and estimates the reservation out of it, which is the conservative default for a model it has no budget fact for; the descriptor is read and reported in status either way, so the gap stays auditable. |
-| `blacklistAutoFoldTools` | `new Set()` | The exception list. Every completed tool batch is foldable by the runtime on its own, without an agent mark; name a tool here and its results stay raw instead. Empty by default, because foldability is not the protection: pins, the fresh tail and the commit guard are, and they apply to a blacklisted tool and an ordinary one alike. Pass a `ReadonlySet<string>`. Blacklisting is not free: the mass stays in the window until a chapter fold takes it, and a closed unit that alone exceeds the chapter cap is reclaimed whole rather than in pieces. |
-| `guidance` | `{ actionResponses: true }` | The persistent acknowledgement returned by a context action. Boolean only, and off means the carrier is absent rather than empty. |
-
-### Migrating from 1.x
-
-Version 2.0 makes the measured deletions explicit instead of accepting options that no longer control the runtime.
-
-- Remove `summarizer` and `summarizeContextSpan`. The model brief generator is deleted; every new fold carries a deterministic or agent-supplied brief. Passing either name is refused with the deletion reason.
-- Remove `guidance.thresholdNotices`. `guidance` now contains only `actionResponses`.
-- The default `thresholds.minTarget` is now `0.20`. A host supplying the whole threshold object keeps its declared value.
-- Existing sessions remain readable. Fold records already carrying model-authored briefs keep those briefs and their provenance on load.
-
-### Projection candidate records
-
-`registerPiFold` returns `{ projectionCandidates }`, a provider a host may call with a Pi context. It emits one record per visible collapsed fold. The fields and current values are:
-
-| Field | Emitted value |
-| --- | --- |
-| `version` | `1` |
-| `key` | `"projection:<fold-id>"` |
-| `kind` | `"projection"` |
-| `domain` | `"system"` |
-| `horizon` | `"working"` |
-| `source_id` | Fold ID. |
-| `source_version` | Exact source SHA-256. |
-| `route` | `{ tool: "pi_fold_context", arguments: { action: "expand", id: <fold-id> } }` |
-| `token_cost` | Estimated replacement tokens, rounded up with a minimum of 1. |
-| `expansion_cost` | Estimated exact-source tokens, rounded up with a minimum of 1. |
-| `rank` | Zero-based position among emitted candidates. |
-| `score` | Saved-byte fraction, capped at 1. |
-| `raw_score` | Exact bytes saved by the fold. |
-| `confidence` | `"exact"` |
-| `freshness` | `"current"` |
-| `locked_owner` | `false` |
-| `collapse_key` | `"projection:<fold-id>"` |
-| `generator` | `"projection-model"`, `"projection-deterministic"`, or `"projection-supplied"` from brief provenance. |
-| `generator_version` | `"memory-slate-generators-v2"` |
-| `recency` | `null` |
-
-## Status
-
-Stable. The option surface is settled and the append-only projection invariant is enforced by the gate suite, which is the contract: `tests/verify.mjs` covers the runtime and `scripts/verify_pi_context_experiment.mjs` covers the experiment harness. New behavior lands with its gate.
-
-## Development
+Two recent lines inform this one. Lossless hierarchical compaction, exemplified by LCM ([arXiv:2605.04050](https://arxiv.org/abs/2605.04050)) with a walkthrough at [losslesscontext.ai](https://www.losslesscontext.ai/), builds a summary DAG over older messages with lossless pointers to every original. Agentic context management, exemplified by [Letta](https://github.com/letta-ai/letta) and by [arXiv:2510.11967](https://arxiv.org/abs/2510.11967), gives the agent tools over its own memory. See also [arXiv:2607.00692](https://arxiv.org/abs/2607.00692).
 
-```sh
-npm ci
-npm run lint
-npm test
-node scripts/verify_pi_context_experiment.mjs
-```
+What this one adds is an efficiency result, arrived at by running the mechanism repeatedly and following the bill. Rotation is the part every approach here shares. The cost sits in where the rewrites land, because a prefix cache charges by position rather than by size, so each fold is staged as a mark that moves nothing and one commit spends the cache for all of them at once. The same campaigns settled the recommended shape: deterministic briefs paired with a durable memory store, which held recall at the lowest cost and wall clock measured. Agent control over boundaries and brief text stays available, and the invitation to use it ships off.