Nine findings from a review of the phase-14 stack. The one about a retry
withdrawing a memory is not a bug — a memory anchored to a node describes that
node, and it goes when the node goes. The root is the exception, and it is the
only one: migration 62 parked every memory written before memories had
coordinates on depth 0, so withdrawing the opening node would retire a whole
bank nobody attached there. A memory with no source range covers no story and
now stays; a summary that genuinely ends there is still withdrawn.
The rest are repairs.
* The adventure list quoted whichever attempt was written last rather than the
one the story tells, so switching back left the index disagreeing with the
page.
* A v1 import gave a typed memory no depth, rebuilding the NULL the migration
exists to remove — invisible until the imported adventure forked.
* The action cap counted a v1 file's turns, and a turn expands into a row per
saved attempt, so a file inside the cap could write a multiple of it.
* Forking a live node on a borrowed ancestor promoted a sibling on a branch the
caller never named. It is a branch switch, and now says so.
* Switching attempts left the state, status and memory panels reading the
previous take: the story does not change length, so nothing keyed on its
length noticed. Same class as the branch-switch bug this phase already fixed.
* A retry after switching back numbered the new attempt into the middle of the
group instead of the end.
* The cursor backfill numbered every action in the table once per adventure;
correlated to the adventure being updated, it is an index lookup instead.
* Renaming a branch answered own_actions=0.
And one behaviour change recorded rather than repaired: script-visible history
and actionCount no longer count blank-text rows. That is the right shape and
there is no reading compatible with both, so plan/14 says so.
409 tests.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
A hand-written memory used to carry a NULL depth, described in the model as
"belongs to the adventure rather than to a path". That sounds harmless and
is not: a NULL is a coordinate no fork can cap, so a note typed on one line
followed the reader onto branches whose story it never described. It takes
the head now — the story you were reading when you wrote it — and obeys
exactly the rule a summarised memory obeys.
The unanchored escape clause in lineage.Path.clause existed for that single
case and is deleted rather than left unused. Its docstring argued that a
capped depth would drop a typed memory the moment its branch stopped being
the newest entry; anchoring answers the same worry better, because the
memory is not exempt from the path, it is on one.
The drawer now shows the path being read and nothing else, filtered by the
clause retrieval itself uses, so the bank you can see is the bank the model
can see. Nothing is stranded: a memory lives on a branch, switching to that
branch shows it, and deleting the branch deletes it. Pinning decides order,
the path decides existence.
Migration 62 lands existing NULL-depth memories at depth 0 of their branch
rather than at the tip. 0 is at or before every fork point, so every memory
stays visible from exactly the paths it is visible from today — nobody's
bank loses a row on deploy. The tip is the tidier-sounding choice and would
have emptied them out of every branch forked earlier than they were typed.
This supersedes the on_path flag and the "another branch" badge from
earlier today; anchoring makes them redundant, and they are removed.
Four tests changed because they asserted the old contract, not because
they broke. The one worth reading is the pair replacing
test_a_hand_written_memory_is_not_lost_at_the_first_fork: typed on shared
trunk it still survives a fork, and typed on ground the fork never
travelled it no longer follows you.
402 tests. Verified on tools/branch_fixture.py: each branch's drawer holds
its own memory and not the other's.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
Retrieval has been path-scoped since SP3: a memory on a branch this story
never travelled is never sent to the AI. The drawer listed every memory
alike, so on a fork you read "Fell down the cellar stairs; badly hurt" and
reasonably concluded the model knew it. It does not. That is worse than
either hiding the row or retrieving it — it is the screen claiming
something the engine contradicts.
Hiding them is not the answer either; that was the original comment's
point, and it stands. A memory nobody can list is a memory nobody can
delete, in a phase whose rule is that nothing is removed automatically.
So the whole bank still lists, and the rows off the current path are set
back, dashed, and labelled "another branch". MemoryOut.on_path carries it,
computed from the predicate retrieval itself uses rather than a second
spelling of the same idea — two spellings drift, and the failure mode here
is a badge that says the opposite of what the model gets. Pinning does not
override it: the path clause runs before pinning is considered.
The relationship is asymmetric and there is now a test that says so. A fork
borrows its ancestors, so a memory written on the parent is on the fork's
path too; the reverse never is. Worth pinning before somebody makes it
symmetric on the grounds that it looks wrong.
399 tests, three new, egress ceilings intact — the flag costs one id-only
query. Verified in a browser on tools/branch_fixture.py.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
A branch switch does not change the length of the story. It changes which
story it is. Four panels keyed on actions.length and so could not tell the
difference: the Branches panel drew one branch while the reader was already
on a second, Insights showed the prompt built for the path just left, the
script-state drawer kept the other line's numbers, and the Memory Bank did
not notice a deleted branch taking its memories with it.
Only the world-state drawer was right, and only because it happened to
carry stateKey already. They all key on the pair now, and deleting a branch
bumps it too — that is the one operation that changes what is stored
without a turn being played and without the story on the current path
moving by a single action.
tools/branch_fixture.py is the thing that could show it. The stress
fixture's world state is empty, so it cannot answer whether a switch puts
the scoreboard back, and its story is one branch. This builds a small
bootable adventure with a stat schema, a gold script, two takes on one turn
that differ by 35 hit points, a fork, and a memory on each side — with both
branches the same length on purpose, because equal length is precisely the
case a length-based key cannot see.
Verified in a browser with the drawer open: hp 60 to 95 and back, the bar
redrawn, the story swapped to the other take, Insights carrying the scratch
and not the beating. The Memory Bank deliberately does not change on a
switch: the drawer is adventure-wide so a memory is always findable to
delete, and retrieval is the path-scoped half.
396 tests, build clean, no new lint.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
The pager could only step between attempts, and stepping has nothing to say
about the thing the tree exists for: taking a path the story moved past and
keeping both. Every attempt has been its own node since SP4, so a chip is a
node now, and "take this path" forks — or simply switches, when the turn is
still the tip and its attempts are leaves nobody has built on.
Beside it, a Branches panel: every line the story has taken, with where each
left its parent and where it ends, and switch, rename and delete-with-confirm.
It sits with Plot/Memory/Scripts/Insights rather than inventing a new place to
put a rail. An unnamed branch is drawn from its fork depth, never from its
position in the list — a position shifts the moment a branch above it goes.
A spatial per-node map was considered and deliberately not built. At the size
this has to be verified against it is a second windowing problem, and it can be
added later without a new endpoint, since the rail and a map read the same
GET /branches. VariantOut grows an id because a fork is addressed by the node
being taken, not by an ordinal in a group that renumbers.
Driven by hand against the 602-action fixture, which found one bug that no test
could: the panel refreshed on actions.length, and a fork swaps a 60-action
window for another 60-action window, so it went on drawing a one-branch tree
while the story was already on the second. It keys off the counter adoptWindow
bumps now.
The scroll path was driven at the same time — three prepends of ~16,200 px, the
same node holding viewport top 792 to 787, never thrown to the end. That closes
the standing gap in this project. Console clean.
396 tests, build clean, no new lint.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
SP7 needs three branch operations and SP5 built one. Switching exists;
naming and deleting had no column and no route between them.
A name is stored because a player chose it. An unnamed branch keeps NULL
rather than a generated "branch 4" — a generated label is derived, and it
would go stale the moment a branch before it is deleted and the ordinals
shift underneath. The client draws those from the fork depth, which
nothing can shift. The v2 bundle carries the name for the same reason it
carries the fork points and leaves `lineage` out: it is a decision, not
something computed from one.
Delete is what stands between a tree and unbounded growth, since nothing
prunes one on its own. It refuses two branches: the root, which holds the
turns every other branch borrows, and the one being read — including any
branch the head was forked from, which is the same mistake in disguise and
the one that would cascade the head away and leave head_branch_id pointing
at nothing. The nodes, memories and descendants go through the foreign
keys that already cascade.
A cursor standing on a deleted branch is cleared. On Postgres a stale
branch id would simply never resolve; SQLite hands the freed id to the
next fork, and then the anchor resolves onto a branch it has never seen
and calls a stretch of story already summarized.
396 tests, 15 new.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
A bundle had one list and a forked adventure has two stories, so export was
emitting every branch's turns interleaved by index — a mangled story rather
than lost data, and unreachable only because forking has no UI yet.
`ai-dnd-adventure-v2` carries the branches, the depth each one left its
parent at, which attempt at every turn is the story, and what each node
left behind. That last one is not decoration: the after-snapshots are what
a branch switch puts back, and a bundle without them imports a tree nobody
can switch inside.
`app/bundle.py` owns both formats and nothing else knows either. The v1
reader stays — those files are already on people's disks — and it is now
the only place a `variants` array exists anywhere.
The rule the module is built on is that a bundle carries what was chosen
and never what is derived. The head branch, the fork points, the live flags
and the anchors are decisions somebody made. The lineage, the head depth,
the legacy `index` and the variant ordinals are computed from those and are
rebuilt on the way in, because a bundle is a text file anybody can edit and
a derived field shipped beside its source is a chance for the file to
disagree with itself where no read would report it.
`index` is the one that stops being academic here. It agreed with `depth`
until SP5, and this is the first writer that has to fill it for a forked
story, where two branches both hold a node at depth 4. It is allocated one
per turn instead: siblings share it, no two coordinates do.
Everything a hand-edited file can get wrong about the shape of a tree is a
400 raised before the adventure row exists, because a half-applied import
is exactly the failure this phase exists to end — a story that goes quiet.
A file wrong about which attempt is live is corrected rather than refused;
that is an invariant of the database, not of the format.
Measured on the 600-action fixture: 587 kB to 911 kB, and all of the
increase is the outcomes at 489 B a node — the coordinates themselves save
57.5 B a node against the old turn-and-variants shape. Twenty forks add
660 B. 4.3% of the import body cap.
381 tests green, 16 of them new in test_bundle_v2.py. No migration, no
vacuum owed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
SP4 and SP5 are written down: what shipped, what it measured, and the
handful of things worth not rediscovering. The bug table at the top is
scored now that six of its seven rows are gone — and the seventh, the
one-turn holdback, is gone for a different reason than the one written
there, which is the correction that matters most. Editing a summarised
action is still unfixed and now says so; no subphase is scheduled for it.
The Open section closes. `retry_of.index` was the last item and SP4
answered it by reusing the retried node's depth.
Two vacuums are owed, SP1's and SP4's, and neither has deployed. One run
after the SP4 deploy settles both. SP8 grows two more columns to drop and
a caveat about which ones are not dead yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015H5qiyiR7gtFQaoDphHZ3g
Attempts pile up at the tip as leaves and cost nothing. The moment the
player takes the story down one the line has already moved past, the two
futures have to coexist — so `tree.fork` gives that attempt a branch of
its own, forked at the depth just before it, and the line it leaves
keeps every turn it has.
One row is inserted and one row is moved. Nothing is copied: everything
before the fork is borrowed through the lineage cached on the branch
row. Measured on a 40-turn story forked twenty times — 21 branches, 140
rows, an 80-action story — the page load costs 31,652 B against the
31,433 B the same story flat costs, and a branch is 103 B of ancestry.
Nothing derived moves either, and that is the part worth keeping: a
memory hangs off the coordinate the parent's attempt still occupies, and
the lineage caps the parent one depth short of it. The fork simply
cannot see it, so it resummarizes that ground from the text it actually
tells, without a line of bookkeeping.
Two things had to change underneath. A new node's depth now comes from
the tip of its branch rather than from the adventure-wide `index`, which
would have left a hole in a fork's path the width of the other branch.
And undo stops at the fork — the turns before it belong to the branch
this one grew out of.
`GET /branches`, `POST /branches/{id}/switch` and
`POST /actions/{id}/fork` are the endpoints SP7's tree view is drawn on.
365 tests green, 18 of them new in test_branch_forking.py.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Every attempt at a turn is now its own row at the same (branch, depth),
with `live` naming the one the story tells. The JSON repeating group on
`actions.variants` is read one last time, by a migration that writes it
out as the sibling rows it always described, and then goes unread.
The snapshots turn around with it: an action carries the state it left
behind rather than the state it started from, because attempts at one
turn share a starting position and differ exactly in their outcome.
Rolling back is "what the node in front left behind", one lookup on the
path, and it is what undo and retry now both read.
And the memory holdback goes. It existed because retry rewrote a row
under a mark that had already moved past it; a retry writes a sibling
now, and replacing what a coordinate says withdraws what was derived
from it — the same repair undo and delete already made.
The assembled prompt is still stored once per turn: it moves with the
live flag, so a superseded attempt keeps only the few hundred bytes that
were its own. Measured on the 600-action fixture: 700 rows for the same
600-turn story, prompt archive byte-identical at 0.50 MB, index 1.8 kB
and page load 62.7 kB unmoved.
347 tests green. `tests/test_story_tree_baseline.py` and
`tests/test_retry_variants.py` pass unmodified — SP4 was allowed to move
the baseline for the variant-count semantics and did not need to.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The memory bank and the story summary each kept a cursor: how many story
actions they had already covered. A count is a position in a list, and this
list moves — delete an action in front of the mark and every later one slides
down a slot, so the mark now covers one it has never read. All the cursor
bookkeeping existed to patch that up.
Both marks are now (branch_id, depth): the node up to and including which the
work is done. A depth is a coordinate along a path, not an offset into a list,
so nothing in front of it can move it. That deletes rather than rewrites
`position_of_index`, `note_action_removed`, `_rewind_cursors_to_index`,
`prune_dangling_memories` and the every-pass clamp in `run_post_turn`.
A memory hangs off the node its block ends on, so a fork inherits its
ancestors' memories without copying any, and retrieval selects through the
branch clause over the *whole* lineage — recall is long-range by definition and
cannot be windowed. Measured: 1,807 B on a story forked twenty times against
1,823 B on a flat one of the same length.
Migrations 53-56 translate the old counts into nodes. They rewrite `adventures`
and not `actions`, so this one needs no VACUUM FULL.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Trust the statement count, not the stopwatch: this machine's suite timings
drift about 20% between runs, so the branch-read regression is recorded as
201 SELECTs -> 2 rather than as seconds.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The identity map holds weak references. A branch row nobody keeps a strong
reference to is collected between two nodes, so resolving the head inside the
placement loop read it back from the database for every node in the flush: 201
SELECTs on `branches` to write 200 actions, and 36 s -> 45 s on the same 297
tests. Nothing about any result changed, which is why only a stopwatch found
it, and why there is now a test counting the reads.
Also: the two reads left un-pathed on purpose say so where they live —
`max_action_index` allocates the legacy `index` and must stay adventure-wide
or two branches issue the same number, and export is a flat v1 bundle whose
reader has no idea branches exist. And `Adventure.actions` keeps its `index`
ordering, because ordering the collection by depth would not make it a story:
it is every branch's actions, and a path is a selection out of it.
318 tests green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Every read of an action now goes through a single module. `context/lineage.py`
turns a branch's stored lineage into the OR-of-ranges that is "this story", and
history, paging, the newest-action lookups, the index screen and the scripting
history API all select through it. A forgotten clause does not raise — it
quietly assembles a page, or a prompt, out of two different stories — so the
clause lives in one place rather than in a convention.
The read that mattered most was the shortcut: `_from_memory` sliced
`adventure.actions`, which is every branch's actions, not the path. It now cuts
the loaded collection down with the same predicate the SQL uses. Same trap one
layer up, and user-visible: `pipeline._history()` hands user scripts the story,
and was handing them the collection.
Tail reads window the lineage as well as the rows: the newest few entries cover
the context budget, so a story forked twenty times reads its tail with one
clause and costs 1.07x what an unforked story of the same length costs. The
estimate is depth arithmetic, and where a deleted action leaves a gap the read
notices it came up short and widens to the whole ancestry.
Ordering moves from `index` to `depth`, with `id` breaking ties. The two hold
the same numbers until retry stops mutating rows in SP4, but only one of them
is a position along a path.
One thing SP1 did not anticipate: wiring the writers was not enough. From here
a row without a branch is a row no read can see, and "every writer remembers"
has to hold for every fixture, script and test ever written — including the
SP0 baseline, which writes its actions straight to the database and must pass
unmodified. So the session enforces it: `tree.place_new_nodes` runs from
before_flush and places anything unplaced. The call sites keep their explicit
calls, because a node placed at the call site is placed before the code around
it reads it back.
316 tests green: the 297 from SP1, plus 19 in test_branch_clause.py.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Phase 14 SP1. The tree goes into the schema and nothing reads it yet: a
`branches` table, `branch_id`/`depth` on actions and memories, a head pointer
on adventures, migrations 46-52, and a server-side backfill that re-reads every
existing adventure as a tree with one branch. `depth` holds the number `index`
already held, gaps included, so no story changes — a linear story *is* a tree
with one branch, which is what makes the SP0 baseline passing unmodified the
pass condition rather than a hope.
The writer had to come with it. No migration will ever visit a row written
after it ran, so columns backfilled today and populated next subphase would
leave a hole exactly the width of one deploy, and from SP2 on a row without a
branch is a row no read can see. `app/tree.py` owns that: one module, because a
node written without a branch fails by disappearing rather than by raising.
Three things the schema itself insisted on:
- `adventures.head_branch_id` is a plain integer, not a foreign key. Pointing
both ways makes the two tables a cycle create_all cannot order, and its
escape hatch needs an ALTER SQLite does not have. It is a cache, and a head
naming a branch that is gone recovers onto the root.
- `lineage` is NOT NULL, so the backfill inserts `'[]'` and fills it in a
second pass guarded on `json_array_length(lineage) = 0` — not `= '[]'`,
because Postgres `json` has no equality operator.
- SQLite will not drop a column a foreign key names, which is how two existing
tests broke: they simulated an old database by rewinding the stamp while
leaving the new columns in place. Every ADD COLUMN migration is now
idempotent, and `tests/test_tree_migration.py` builds a genuine schema 45 by
rebuilding three tables from frozen DDL so the real ALTERs run.
297 tests green, 14 of them new. `branches` costs 0.1 kB of a 733.5 kB turn;
page load and index are byte-identical to the recorded figures.
The deploy that ships this needs one `VACUUM FULL actions;` on the direct
endpoint afterwards — it rewrites every row.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The stress fixture is sized from production and exists to weigh bytes, so it
leaves every column it does not weigh at its default. Checked against a freshly
built one, that is exactly the set phase 14 has to migrate: state_before and
world_state_before NULL on all 600 rows, no scenario and so no RPG layer, no
adventure scripts, both cursors 0, and a hundred retry histories whose two
attempts carry byte-identical text with variant_index pinned at 0.
That last one matters most. "Which attempt is live?" is the question SP4's
migration answers when it decides which sibling node becomes the head of a
turn, and against that fixture the question had no observable answer -- a
migration that picked wrong would have looked correct.
--rich fills in those columns and no others. A real RPG scenario read from the
seed data rather than invented, with a world state played forward so hp
declines and flags flip; monotonic per-action state snapshots, so a rollback
that does not happen reads as a wrong number instead of as nothing; a gold
script, story cards, non-zero cursors, pinned and forgotten memories; retry
attempts with distinct texts, counts of two and three, and a live attempt that
is often not the last written. Plus a second adventure, because a branch clause
that forgot its adventure still looks right on a database holding one.
The invariant SP4 reads -- text mirrors variants[variant_index] -- is asserted
at build time rather than assumed.
The plain fixture is untouched and re-measured unchanged at 1.8 kB, so the
egress ceilings stay comparable. All six shapes run clean on --rich, including
the turn path with the RPG layer and script pipeline now live. 283 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Phase 14 replaces the action list with a tree, which touches the memory bank,
the context builder, undo, retry and the UI at once. Before any of that moves,
write down what "unaffected" means and prove it holds today.
tests/test_story_tree_baseline.py drives the product over HTTP the way a player
does and asserts only on API responses, never on how anything is stored: the
turn engine, retry and variant switching, undo, paging all the way back to the
start without gaps or repeats, edit/delete, memories, export/import round-trip,
world state. 283 green with it added, from 259.
It must pass unmodified through the schema change, the branch clause and the
move of memories onto nodes. A linear story is a tree with one branch, so those
subphases have no licence to change behaviour, and this is what says so.
The plan itself records four decisions that were open: structural-first with no
runtime flag, legacy columns kept one release, full tree visualisation, and a
v2 bundle with a v1 reader. Plus four traps found by reading rather than
running -- history._from_memory and the scripting pipeline both slice
adventure.actions, which under a tree is every branch rather than the path.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The harness already built a production-shaped 600-action adventure and then
threw it away with the temp file. The one open gap in plan/13 is that nothing
has ever driven the scroll in a browser, and part of why is that there was
never a long adventure to drive it with.
--keep PATH writes the fixture somewhere durable and makes the app able to
serve it. Two edits are needed for that, both of which cost an hour to
rediscover:
- create_all() builds the current schema but leaves the version stamp at its
default, and bootstrap() reads a populated-but-unstamped database as
ancient — it replays every migration against a schema that already has the
columns, and fails on the first.
- the fixture's user is a registered one, but local mode looks for the row
with email IS NULL and is_guest false, so without clearing the email the
app opens on an empty library.
--keep is read before argparse exists, because where the database lives has to
be settled before app.database is imported. That is the same constraint the
AIDND_STRESS_DATABASE_URL block already lives under. SQLite only; combining it
with a Postgres target is rejected rather than half-honoured.
Verified end to end: the fixture boots with no manual step, action_count 600,
a 60-action first payload, and before_id walks back nine more pages to the
start. Nothing about the default path changed; 259 tests pass.
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
The post-vacuum sizes had never been read. They were 144.2 MB - above the
99.6 MB the compression work started from, not the ~53 MB it projected. The
earlier VACUUM FULL ran before migrations 42-45, which then rewrote every row
and refilled the table with dead space that only another VACUUM FULL returns.
Ran it: 144.2 -> 65.0 MB, 79.2 MB reclaimed in 5.5s, health green after.
actions is now 52.1 MB holding 50.4 MB of live bytes. The projection was right;
nothing had reclaimed what the compression freed.
Also records where those bytes are. context_snapshot is 94% of the table at
104 kB a row, and the per-action state columns a tree would sit beside are
rounding errors - which is the number phase 14 wants before it adds more.
And a trap: n_live_tup on the TOAST relation implied half the real bloat. It
is a leftover ANALYZE estimate, stale in the flattering direction right after
a migration. Size from sum(octet_length(col)) instead.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
STATUS was written before the deploy and still asked for a VACUUM that has
since been run. Replaces that with what was actually verified against the
running service, and with the two things a next session would otherwise have
to rediscover.
The post-vacuum sizes were never measured, so the projection stays a
projection; the query to settle it is in the file. And the scroll gap is
promoted from a footnote to the one real open item, because re-reading that
path after shipping found a bug in it -- which is the argument that reading it
again is not the fix.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Prepending older actions changes `actions`, and the bottom-pinning effect
watches `actions`. If nothing had un-pinned the view first, loading earlier
turns scrolled straight past them to the newest one -- the reader asks to go
back and the story throws them forward.
Scrolling up is what normally un-pins, so the case that breaks is the one
where the reader never scrolled: a window short enough to fit the viewport,
where the "N earlier moments" button is the only way back and the button is
what fails. Reading backwards is the opposite of following along, so a prepend
now clears the pin explicitly rather than relying on a scroll having happened.
Found by re-reading this path rather than by running it -- the scroll
behaviour is the one part of the paging work nothing exercises, since the
frontend has no test runner.
Also holds off three seconds after a failed page fetch. Parked near the top,
momentum scrolling calls this dozens of times a second, and against an
endpoint that is failing that is a retry storm rather than a retry.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The backfill issued an UPDATE per action. It runs at container start, before
uvicorn binds the port, against a database at the other end of a network --
944 rows on production, so a thousand sequential round trips standing between
the deploy and its first health check.
One executemany per batch instead. Same rows, same verification, same
transaction; nineteen round trips rather than nine hundred.
Re-verified on Postgres, since executemany binds bytea through a different
psycopg path: 720,864 B of JSON to 204,293 B, every row equal.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Records what the six changes measured, and replaces the pick-up item -- which
was step 6 -- with plan/14, since there is nothing left in 13.
Two things a future session needs and cannot infer from the code. The VACUUM:
migration 43 compresses context_snapshot but Postgres does not return the disk
by itself, so until `VACUUM FULL actions` runs the storage win exists only on
paper and the table is temporarily larger, not smaller. And the gap: nothing
exercises the scroll behaviour in a browser, because the frontend has no test
runner, and prepend-and-restore-scroll is the part most likely to feel wrong
even when it is correct.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Opening a finished adventure fetched every action in one response: 589.5 kB on
production's longest, and nothing about that curve bends on its own, because a
story only ever gets longer. The page load now brings the newest 60 actions
and the reader pages up from there. On the harness's 600-action fixture that
is 606.0 kB down to 62.6 kB, and -- the part that matters -- it no longer
depends on how long the story is.
Paged by anchor, not by offset. `before_id` is the oldest action the caller
holds; the server returns what precedes it. An offset counted back from the
newest would shift every older position the moment a turn lands, which is
exactly when someone is likely to be scrolling, and the reader would get one
action twice and never see another. It also keeps working when the story stops
being a flat list: comparing indices to order a branch survives the story tree,
treating them as positions does not.
`has_more` comes from fetching one row past the window rather than from
counting. A deleted anchor -- undo, mid-scroll -- reports the end rather than
guessing and serving a page the reader already has.
GET /{id}/actions and POST /{id}/undo now return {actions, total, has_more}
instead of a bare list. Undo is the action most likely to be repeated several
times running, so having it re-fetch the whole story would have undone the
paging on the worst case.
The adventure payload gets its window through set_committed_value rather than
by assignment: the actions relationship cascades delete-orphan, so assigning a
60-item list to it would delete everything outside the window on the next
flush.
In Play.jsx the prepend is followed by a useLayoutEffect that restores the
scroll position, before paint, so the story does not jump. Loading starts 400px
from the top rather than at it, guarded by a ref because scroll fires far
faster than React re-renders. There is a button as well as the scroll trigger:
on a short viewport the transcript may not be tall enough to scroll at all, and
a reader who cannot scroll must still be able to reach the beginning.
Verified against a running backend and a 220-action adventure: the page load
returns 60 of 220 ending on the newest, and walking back from an anchor returns
exactly the actions before it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
One column is 89% of the database and the free tier allows 512 MB. Reads were
already solved -- the column is deferred, so a page load never touches it and
one screen fetches one row at a time -- but nothing had costed storage, and
storage is the constraint with a cliff: 99.6 MB used, ~94 kB of disk per
action, so the ceiling arrives around 5,400 actions and 944 are stored.
Postgres already compresses it and only gets 1.7x. pglz is tuned for fast
decompression of data a query might filter on, and nothing has ever filtered
on an assembled prompt -- it is written once and read whole, rarely, by the
Insights viewer. zlib gets 3.5x on the same text for a decompress on a request
that already made an LLM call.
Done as a TypeDecorator rather than a second column, so every call site still
writes a dict and reads a dict back, and deferred/undefer/load_only keep
naming the same attribute. Only the storage format moves.
Migrations 43-45: add the bytea, convert into it, drop the original, rename.
The backfill is the one destructive step in the file -- 44 removes the only
other copy -- so it decompresses every row and compares it against what went
in, and a row that fails aborts the run. The whole loop is one transaction, so
an abort rolls the DROP back and the prompts are still there.
Verified on real Postgres, replaying 43-45 from a pre-43 schema on a throwaway
Neon database: 720,864 B of JSON became 204,293 B of bytea, 3.53x, the column
came out named context_snapshot, every snapshot compared equal and the one
NULL stayed NULL.
Postgres does not return the disk by itself: DROP COLUMN only marks the column
gone and the backfill leaves a dead tuple per row, so the table peaks near
twice its size before settling. The deploy needs one VACUUM FULL to collect
it; the migration comment says so.
The egress fixture's snapshots are prose now rather than "x" * 20_000, and the
prose generator moved to tools/fakeprose.py so the harness and the tests share
one definition. A repeated character compresses a thousandfold: against the
old fixture a compressed column looked free and the byte ceilings would have
been guarding nothing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
deferred=True keeps the four heavy Action columns out of a bulk read, but it
makes narrowness the thing a future column has to remember to ask for -- and
both egress blowouts this project has had were a column nobody remembered.
Listing what each list response renders inverts the default: a new column
costs nothing on these paths until someone adds it to the tuple.
The adventures index was not merely a future risk. It loaded whole Adventure
entities to render a title, a stamp and a snippet, and an Adventure carries
script_state, world_state, placeholders, story_summary, memory, authors_note
and ai_instructions -- ~15 kB a row in production, none of it on that screen,
all of it fetched once per adventure on every index load. Measured on six
adventures with 78 kB of body each: 469.7 kB entity-loaded against 318 B
projected.
The memories drawer stops walking adventure.memories. The walk is what
retrieval used to do and the reason a turn cost megabytes; a relationship load
takes whole entities, so it picks up whatever the model happens to grow.
Nothing changes today -- embedding_blob is already deferred -- which is the
point.
world_delta stays on the action list because ActionOut.world_changes is
computed from it. Leaving it off would not save the bytes, it would spend them
one row at a time as a lazy load.
Two tests cover the index: one asserts the listing query names none of the
body columns, one puts a byte ceiling on six adventures carrying 80 kB apiece.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
test_egress.py asserted which columns a statement names, which is the shape
both of this project's egress blowouts took. It would all still pass if a
response grew tenfold within the columns it is allowed to read -- and a story
that keeps getting longer does exactly that. Production's longest adventure is
607 actions where the plan assumed 200.
So dbmeter, which was built to be importable from tests and was not yet used
by any, now backs four byte ceilings: the page load, the action list, and one
action's snapshot fetched on demand. Budgets are per action rather than
absolute, so they mean the same thing whatever size the fixture is set to, and
generous -- 3 kB against a real 994 B. They are there to catch an order of
magnitude, not to freeze a byte count.
The fourth test is the one that keeps the other three honest. A ceiling proves
nothing unless the thing it excludes would breach it, so it undefers the
snapshot on purpose and asserts the same twelve rows cost more than ten times
the budget. If the fixture ever shrinks below the point where that holds, that
test fails rather than the ceilings quietly passing on nothing.
Meter grows detach() and a context manager. A script exits and takes the
wrapping with it; a test does not, and one test leaving the shared engine
metered would charge bytes to a scope nobody opened.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The old fixture was wrong in both directions at once and happened to land near
the right total. Actions were modelled at ~2.1 KB against a real 886 B of
text, and adventures at 200 actions against a real 607. Width flattered,
length did not, and length is what a page load pays for.
Re-sized from the 2026-08-17 measurements: 600 actions, 1700 B of narration
alternating with a one-line player input, 232 KB of context_snapshot a row.
The page-load shape now reports 606.0 kB against the 589.5 kB measured on
production's longest adventure -- 2.8% out, where the old defaults were 28%
out on a story a third of the length.
Filler text is now generated word by word instead of one sentence repeated.
That matters for what comes next: the repeated string compresses 313x and the
generated prose 3.7x, so any compression ratio measured against the old
fixture would have been fiction, and shrinking context_snapshot is the open
question it exists to answer.
context_snapshot also gains a flag of its own rather than being hardcoded, and
the 74 KB figure in the comment -- inherited from models.py -- is corrected:
the real column averages 163 KB a row across the table.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Migration 38 left memories.embedding in place so a rollback could still find
the vectors. Production has since been verified reading from embedding_blob,
so migration 42 drops it: 4 MB of a 99.6 MB database holding nothing anyone
reads.
Removing it surfaced a live bug. Changing your embedding model is supposed to
throw the bank's vectors away and let the post-turn pass rebuild them, because
two models' vectors are not comparable. The settings route did that by nulling
memories.embedding -- correct until 38 moved the vectors, after which it
cleared the dead column and left the blob intact with `embedded` still true.
_embed_pending filters on `embedded IS FALSE`, so it never saw those rows and
the bank went on ranking against the old model's vectors permanently.
Nothing would have reported it. cosine returns 0.0 on a width mismatch, so a
different-width model scores every memory zero and retrieval returns whichever
rows happen to sort first; a same-width model scores plausible garbage.
The bulk clear now sets both columns. It stays a bulk UPDATE rather than going
through set_vector -- loading the rows is the cost that whole path exists to
avoid -- so set_vector's docstring now names it as the one caller that
legitimately writes those columns by hand. No cache invalidation is added:
clearing `embedded` drops the rows out of the catalogue query, and set_vector
evicts each entry as the re-embed puts it back.
test_embedding_blob.py now rebuilds the pre-38 schema by hand where it tests
the backfill, since create_all no longer produces the column it converts from,
and asserts 42 removes it at the end of a full bootstrap -- 38 reads that
column and 42 drops it, so an upgrade that reordered them would arrive with an
empty bank.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
The previous commit recorded the service as suspended. It is not. Render
appended a suffix to the `ai-dnd` service name in render.yaml, and plain
ai-dnd.onrender.com is a different, suspended service whose 503 page reads
"suspended by its owner" -- indistinguishable from this deploy being down
unless you notice the host is wrong.
The real host answers {"ok":true} on /api/health in 0.6s, mints a guest,
returns an empty adventure list for that guest and serves the starter
scenarios. The authoritative link is the one docs/index.html points at, not
the service name in the blueprint.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
Everything measured so far ran on SQLite against a synthetic fixture, so two
claims were still on trust: that migration 38 spells BYTEA correctly for a real
server, and that the byte figures survive psycopg's encodings.
Both hold. Production reads schema_version 41 with embedding_blob bytea and
embedded boolean present, the backfill is complete at 134/134, and the packed
vectors are 5.04x smaller than the JSON on real data -- 30,971 to 6,144 bytes a
memory, as predicted. stress_session now takes AIDND_STRESS_DATABASE_URL, and
against a throwaway Neon database every shape lands within 0.5% of the SQLite
run: the warm turn is 121.1 kB against 122.3, with memories down to 1.7 kB of
it.
The harness writes, so it refuses any target whose name does not say stress or
scratch -- pointed at the production database it stops rather than seeding it
with a fake user and 200 fake turns. It also empties a Postgres target before
building, which a fresh SQLite temp file never needed.
Two corrections fall out, both recorded in plan/13. The page-load model has the
wrong shape: real actions are half the fixture's weight but real stories run to
607 actions, not 200, so the worst real page load is 589.5 kB. And the decision
to leave context_snapshot in the database costed egress but never storage --
it is 88.9 MB of a 99.6 MB database against a 512 MB free tier, which is the
ceiling this deploy will hit first.
Measured with counts and octet_length sums only. No user content was read.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017Dvvqn9ZDR4ixeFPHNbww7
plan/STATUS.md: what the last session changed, what to pick up next, and the
handful of things that were learned the hard way and would otherwise have to
be rediscovered. Linked from the overview, which now also lists plans 13 and
14 alongside the earlier phases.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
Retrieval walked adventure.memories, so every turn loaded every row of the
bank with its vector attached -- 3.1 MB, 96% of everything a turn read. It
now asks SQL which memories are in play (an id and a flag per row), ranks
against vectors held in process, and fetches text only for the five it picks.
Two more callers were doing the same thing and the production SQL could not
see them: _evict_over_capacity walked the bank to count it, and _embed_pending
walked it to find the rows with no vector. Both are counts and filters the
database can do without sending anything back.
one turn 3,258.7 kB -> 723.4 kB cold, 122.3 kB warm
run_post_turn 3,139.1 kB -> 0.7 kB
Insights 3,223.7 kB -> 117.9 kB
Memories drawer ~3.1 MB -> 23.6 kB
A played turn is turn plus post-turn work: 6.4 MB down to 123 kB.
The cache needs no invalidation callbacks, which is what makes it safe. A
vector can only change through set_vector, which drops that one entry;
anything that removes a memory from play leaves the catalogue query, and
entries missing from the catalogue are dropped on the next read. So eviction,
deletion and pruning have nothing to remember to call.
memories.embedded joins the blob, for the same reason actions.variant_count
sits beside actions.variants: with the vector deferred, every "is this
embedded?" check would otherwise be a 6 KB lazy load, once per row.
Capacity drops 200 -> 80, on retrieval quality as much as cost -- ranking two
hundred memories to pick five buries the five. Eviction was measured at scale
first: trimming 100 to 80 costs 0.8 kB and reads no vectors.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
A 1536-dimension vector spelled out as JSON decimals is ~31 KB. The same
numbers packed as float32 are 6,144 bytes, and the whole bank is read on
every turn, so those bytes are paid over and over.
It is a format change, not a precision trade: the endpoints compute in
float32 and render that into JSON, so converting back recovers the original
bits exactly. Nothing is re-embedded and no API call is made -- migration 38
is a pure repack of what is already stored.
Unlike migrations 36 and 37 this backfill cannot be expressed in portable
SQL, so it comes through Python, batched, and pays a one-time read of every
vector to stop paying three megabytes a turn.
The JSON column stays, still written through set_vector, so a rollback finds
the vectors intact. Reading from the blob comes next; a follow-up migration
drops the old column once that is verified.
Migration SQL can now be a {dialect: sql} map -- BLOB and BYTEA have no
common spelling, and every Postgres deploy replays this one.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
Both egress blowouts this project has had were one query fetching a column
nobody read, and a statement count would have shown nothing wrong in either.
So the meter counts bytes, at the DBAPI cursor -- everything that crosses
that line crossed the wire.
tools/stress_session.py drives a production-shaped adventure through the real
routes with only the network faked. It reproduces both figures measured
directly on production: 426.7 kB for a 200-action page load against 423 KB,
and 3,258.7 kB for one turn against 3,153 kB.
The memory bank is on by default, which is the whole point -- the previous
harness ran without an embedding model, so retrieval returned early and the
heaviest read in a turn never happened. --no-embeddings reproduces that
deliberately, and the gap is 29x.
It also turned up two callers the production SQL could not see: run_post_turn
walks the whole bank again every turn, and Insights pays for it a third time.
A played turn costs ~6.4 MB, not 3.2.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
Two plan documents from designing the branching story tree. The tree work
turned up that memory retrieval fetches the whole bank's embeddings every
turn -- 96% of a turn's database traffic -- and that is the one read a tree
cannot window, so it lands first.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
Both composers are position:sticky;bottom:0, which pins them to the bottom
of the *layout* viewport. A phone keyboard does not shrink that viewport --
it only covers it -- so the first tap on the input left the box sitting
behind the keyboard. Tapping back and focusing again appeared to work only
because the page had been scrolled during the first attempt, so the
browser's scroll-into-view landed somewhere else.
Two engines, two halves:
- Chromium honours interactive-widget=resizes-content on the viewport meta,
which makes the layout viewport shrink when the keyboard opens, so 100dvh
and sticky bottoms account for it on their own.
- iOS Safari ignores that flag and only shrinks the visual viewport, so
measure the covered height from visualViewport and publish it as
--kb-inset; the composers offset their sticky bottom by it. On browsers
that already resized the layout viewport the measurement is ~0, so the
same rule is a no-op there instead of a double lift.
The pages also gain a matching bottom margin so the last beat can still be
scrolled clear of a lifted composer.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011BC7Dsg4gSQm3NLcZcG5SK
The bar ran in prompt order while the legend below it ranked by share, so
reading one against the other meant matching colours back and forth. Sort
the segments the same way the legend already sorts, so the widest slice is
always the leftmost and the two read as one list. The sections further
down the panel keep prompt order — that is the order the model sees.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011BC7Dsg4gSQm3NLcZcG5SK
The panel reported one number for the whole prompt and one flat fill bar,
so "3412 / 4000 tokens" never said which section was spending it. Replace
the fill with a stacked bar — one segment per context section, scaled to
the budget so the leftover width is the remaining headroom — plus a legend
ordered biggest-first with each section's share and token count. Hovering
either the bar or a legend row dims the rest; clicking a row jumps to that
section's text below, and each section header now carries its own share.
Section colours move out of CSS into a single SECTION_COLORS map applied
inline, so the bar, the legend and the section headers cannot drift apart,
and related sections keep their hue family without sharing an exact shade
(two identical colours read as one segment in a stacked bar). The narrow
side panel only fits one legend column, so the long tail folds behind a
"+N smaller sections" toggle.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011BC7Dsg4gSQm3NLcZcG5SK
The app had no cleanup of any kind: in multi-user mode every first visit
mints a users row, so the demo has been accumulating one permanent account
per visitor along with everything they generated.
cleanup.py sweeps guests idle for AIDND_GUEST_RETENTION_DAYS (default 5),
once at startup and then every few hours. Startup is the load-bearing
trigger — the free tier sleeps after ~15 minutes, so a long timer rarely
gets to fire.
Idle is COALESCE(last_seen_at, created_at), not last_seen_at: _touch only
writes that column hourly, and a guest minted by /auth/me has it NULL until
its second request, so the simpler query would have deleted brand-new
visitors mid-session.
It's one Core DELETE rather than db.delete(user), which would SELECT every
adventure, action and memory into Python purely to delete them — the same
egress pattern as the 189x fix. Every FK from users down is ON DELETE
CASCADE, so the database does the whole graph and returns a count.
The filter requires is_guest AND email IS NULL, so registered users (who
upgrade in place) and local mode's implicit user are both out of reach, and
is_public is output-only so a guest can never own content another user can
see. Session cookies have no expiry and can outlive a swept row; that path
401s and the frontend's existing retry re-mints a session.
Guests are told: /auth/me serves guest_retention_days and the signup modal
states the window, sourced from the server so it can't drift from what is
enforced.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
The per-IP rate limits could be bypassed entirely: uvicorn ran with
--forwarded-allow-ips "*", which trusts the leftmost X-Forwarded-For value
(client-controlled), and Render forwards the inbound header rather than
stripping it. Rotating the header handed out a fresh rate-limit bucket per
request, so the login/register limit (10/5min) and guest-minting limit
(30/5min) were no throttle at all — unbounded password guessing and guest-row
creation. Confirmed live: fixed IP -> 429 after 10; rotating spoofed header ->
no 429 across 14 attempts.
Two-layer fix:
- limits._client_ip now derives the client IP from the hop the trusted edge
appends (rightmost of X-Forwarded-For), which a client can't spoof past;
tunable via AIDND_TRUSTED_PROXY_HOPS. Dropped --forwarded-allow-ips "*".
- New per-account login throttle (email-keyed, 8 fails / 15 min, cleared on
success): stops distributed guessing against one account that a per-IP limit
can't, since it can't be diluted across many source addresses.
Also close an SSRF on the BYOK endpoint_url (hosted mode only): the connection
test and turn/chat streams now refuse a URL that resolves to a non-public
address (private/loopback/link-local metadata/reserved), checked at request
time so it resists a DNS record flipping to a private IP. No-op locally, where
reaching localhost Ollama is intended.
Tests: test_ratelimit_hardening.py (8), test_netguard.py (13). 172 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015CYEJKobJ2Re4Dv7qUoSA7
The hero linked at the Render hostname directly. Pointing it at the project
page keeps the deploy URL out of the README and folds in the "prefer a tour
first?" line, which pointed at the same place.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PoBAfwRzHozF2bhZumxjPk
The guide linked straight to the Render hostname in three places. That URL is
an artifact of where it happens to be hosted; the project page is the stable
entry point and already carries the demo link.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PoBAfwRzHozF2bhZumxjPk
The README says what the project does; nothing said why any of it works the
way it does. This adds design notes written to be read end to end: each
section states the decision, the reasoning, and what it cost.
docs/GUIDE.md is the readable source. docs/guide.html is the same material as
a self-contained reading page for the project site — no webfonts, no scripts
beyond a progress rail, so it also works saved to disk and opened offline.
Covers the context budget allocator, the propose-and-referee world-state
engine, the measured length-hint result, the memory bank's settled-action and
cursor rules, the two coordinate systems behind the summarization bugs, the
retry variant machinery, the egress fix, and the demo-key pinning. Closes with
the measured numbers, the known limitations, and a pointer to the cleanup
backlog in self-review.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PoBAfwRzHozF2bhZumxjPk
The demo sleeps on Render's free tier, so a cold link looks broken to
anyone who won't wait 30-60s. A static page on GitHub Pages is never
asleep: it shows the screenshots immediately and sets the expectation
before the visitor clicks through to the demo.
Served from main:/docs, reusing the screenshots already committed there.
Palette and type match the app so the two read as one product.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015sygnUWH8WnaoKPDp7hXM1
The README claimed screenshots were "coming" and stopped describing the
project around phase 10, so the world-state engine, the retry/undo state
revert and the egress work — the most interesting parts — were invisible.
- Screenshots of the play screen, Insights, the schema editor and the
script editor, captured from the running app.
- Document the world-state engine, undo/retry rewind, and the two measured
performance fixes; correct the context assembly order to match
context/builder.py; drop the stale "phases 1-6 complete" note.
- Add CI (backend pytest, frontend lint + build, Docker image build) and
its badge. Nothing ran the 151 tests but me.
- Move CODE_REVIEW_FINDINGS.md to docs/self-review.md and say up front
that every correctness finding is resolved.
- Drop a dead import that the newly-wired lint flagged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015sygnUWH8WnaoKPDp7hXM1
