What the audit is worth keeping, and where

Five passes produced a working document that said on its first line it was temporary. This is it being spent rather than abandoned. CLAUDE.md gains eleven paragraphs, each a thing that had shipped looking correct: a handler bound to a shared variable rather than its own socket, a script the base template already loads being loaded again, a control that stays clickable while it awaits permission, "is this name taken?" asked about visibility instead of ownership, root running a file the service account can write, sourcing anything under $PREFIX, a read-only command name that is not a read-only command, 0.0.0.0 being this machine, a folder that is not a label, a file that is not deleted by the row that named it, and a measuring harness that measured an unstyled page and reported a dramatic finding that was entirely an artefact. PLAN.md carries the seven things the audit found and deliberately did not fix, each with why: they change what something does rather than fix what it claims to do, which is not an audit's job. deploy/README.md says why root runs a copy, and that a host installed before this keeps the old wiring until the installer is re-run -- the button cannot fix it, because the button runs the old unit. docs/notes/audit-0.9.md is deleted, having been all three. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-07 15:25:44 +02:00
parent 0514568df0
commit e7294e5b20
3 changed files with 135 additions and 423 deletions
@@ -20,7 +20,9 @@ lembas info                         # paths + counts, useful when confused
 lembas secret-key                   # generate LEMBAS_SECRET_KEY
 lembas create-admin                 # create or promote an admin
-pytest                              # 2088 tests, ~2min
+pytest                              # 2283 tests, ~2.5min
 pytest -m "not slow"                # skips the 365 that stand up a real
                                    # server, shell, PTY or git repository
                                    # PLAN.md tracks what is and is not built
 ruff check .                        # lint (line length 100)
 python scripts/build_artwork.py     # regenerate artwork (SVG + PWA icons;
@@ -221,6 +223,10 @@ touching the code it names -- these are the same notes, not a summary.
 - `docs/notes/branding.md` -- the branding snapshot and why it is a Jinja global,
  where the instance name went and how an upgrade keeps it, how a custom theme
  inherits through `data-base`, and why `/branding.css` is a route.
 - `docs/notes/release-checklist.md` -- the manual pass before a release. Not a
  corner of the code but the opposite: the things a machine cannot test, which
  is everything needing a real endpoint, a real machine, real hardware or a real
  pair of eyes. Read it when cutting a release, not while writing code.
 - `docs/notes/subagents.md` -- the hidden chat a helper runs in, why `unattended`
  is a column and not a kind, the two halves that stop a helper stalling on a
  card nobody can see, what it may run and why Auto is never inherited, and where
@@ -1645,6 +1651,107 @@ that the right one works, because only the second half would have passed
 throughout. When adding a control that writes, check the verb against the route,
 and assert on the row rather than on the response.
 **A handler bound to a shared variable is bound to the wrong thing.**
 `terminal.js`'s `socket.onclose` set the module-level `socket` to null, and
 `close()` *queues* its event rather than firing it -- so on either reconnect
 path the old socket's close arrived **after** a new one had been assigned and
 nulled the live one. Output kept arriving, because `onmessage` is bound to the
 object; every send gates on the variable, so each keystroke was dropped and no
 resize was ever sent again, on a panel that looked perfectly healthy. It then
 said "Disconnected" about a shell that had just reconnected, because
 `closedOnPurpose` had already been cleared for the new connection. Each handler
 captures its own socket in a local and returns early when `socket !== opened`.
 **A script the base template loads must not be loaded again by a page.**
 `messages/index.html` listed `composer.js` and `commands.js` in its `scripts`
 block and `base.html` already loads both, so that screen ran each **twice**.
 Each is an IIFE with its own state, and `stopPropagation()` does not stop a
 second listener already bound to the same node: two composer menus stacked, one
 Enter running a command twice, and `Alt+B`/`Alt+E`/`Alt+T`/`Alt+I` toggling
 their panel twice, which is to say doing nothing at all. Nothing about that
 looks like a double load. `tests/test_ui_js.py` sweeps every template against
 `base.html`'s list.
 **Leave the state a control was in before you await.** `audio.js` moved the
 microphone out of `idle` only once `getUserMedia` had resolved, and the
 browser's permission prompt sits in between -- so every click while it was up
 opened another stream and another recorder, of which only the last was ever
 stopped. The browser's recording indicator then stayed on until the tab closed.
 `working` already existed and was already styled; it was the state the button
 was missing.
 **"Is this name taken?" is a question about ownership, not visibility.**
 `skills.create` checked with `by_name`, which is scoped through
 `sharing.visible_to` -- owned *or* shared -- against a `(owner_id, name)`
 constraint. So a skill somebody shared with you took that name out of your
 library, and refused yours with "edit it instead", naming a row you cannot edit
 because sharing grants reading only. `by_name` is right for `skill_get` and
 `skill_edit`, where a shared skill is what the model is reaching for;
 `owned_by_name` is the uniqueness question. `documents.create_base` next door
 had always asked it correctly.
 **Root must not run a file the service account can write.** The update helper's
 unit pointed `ExecStart` inside the checkout, which `install.sh` clones *as* the
 service user -- so anything able to write as that account could rewrite the
 script, create the request file it also owns, and be root. Worse, no compromise
 was needed: an update pulls new code as that user and root then runs whatever it
 fetched, so control of the branch was control of root. `install.sh` puts a
 root-owned copy at `/usr/local/lib/lembas/update.sh`. The cost -- changing
 `update.sh` needs the installer re-run -- is the right one, and `update.sh`
 detects both the stale copy and the old wiring. **The test that existed asserted
 the vulnerable `ExecStart` line verbatim**, and had pinned it in place.
 **Nothing under `$PREFIX` may be sourced.** `update.sh` did `. "$PREFIX/.deploy-env"`.
 The file is root-owned, having been written with `sudo tee` -- but `$PREFIX` is
 the service account's own directory at mode 755, and write permission on a
 *directory* is all it takes to unlink a file and put another there. The two
 values it wants are parsed with patterns now. The test asserts nothing under
 `$PREFIX` is sourced rather than naming `.deploy-env`, because the next file
 read from there would have the same problem.
 **A read-only *name* is not a read-only *command*.** `policy._UNSAFE` stops a
 line being composed out of two commands and says nothing about one command that
 composes another itself -- and the obvious read-only tools do: `find -exec` runs
 a program, `-fprintf` writes a file, `-delete` removes one, `rg --pre` runs a
 preprocessor per file. None needs a character `_UNSAFE` refuses. `find *` was on
 `SAFE_COMMANDS`, which is what a **subagent** is pinned to in every mode,
 unattended, with no card anybody could approve. `policy._ACTION` refuses those
 flags in `subject()` rather than the list being trimmed: a glob cannot say "and
 no dangerous flags", and "this one looks read-only" is exactly the reasoning
 that put `find` there.
 **`0.0.0.0` is this machine.** `hosts._literal` answered from `is_loopback`, and
 the unspecified addresses are not loopback -- so it returned a *decided* False,
 which also short-circuited `resolves_here` and skipped the DNS half. Connecting
 to `0.0.0.0` or `::` goes to loopback on Linux, so an SSH profile pointed there
 reached this host's own sshd with the guard still reading "off".
 **A folder is not a label.** `effective_system_prompt` walks up from the chat
 through its folder and that folder's parents, so a chat filed in somebody else's
 folder takes their system prompt. `_new_chat` resolved the folder, discarded it
 when it was not the caller's -- and then stored the raw `folder_id` anyway, so
 the ownership check governed which *seeds* were applied and not where the chat
 went.
 **A file is not deleted by the row that named it.** `db.delete(chat)` cascades
 to messages and attachment *rows* and leaves every file on disk, with nothing
 that will ever look at them again -- `sweep_orphans` only considers uploads that
 were never attached. `chat_service.delete_chats` is the one way to delete a chat
 and unlinks first, while the rows still say which files they are. Underneath it,
 `files.claim` bound `message_id` and never `chat_id`, so anything picked on the
 *new-chat* screen kept an empty `chat_id` for life -- and six readers filter on
 that column, so those files were unnamed in the prompt, unopenable in the canvas
 and invisible to the one caller the cleanup had.
 **Drive the JavaScript, and check the harness before believing it.** Three of
 the four bugs the testing pass found came from a Node DOM stub. The fourth
 lesson is about the instrument: a headless-Chromium harness that renders a page
 through `TestClient` and rewrites its asset URLs reported **all five tab panels
 visible at once** -- because the templates emit absolute `http://testserver/static/…`
 URLs via `url_for`, not `/static/…`, so the rewrite matched nothing and it was
 measuring an unstyled document. A dramatic finding that was entirely an
 artefact. Any such harness must fail loudly on an unrewritten asset URL.
 ## Releasing, and the changelog
 **Every version bump gets a `CHANGELOG.md` entry, in the same commit.** Not
@@ -95,19 +95,40 @@ The button is opt-in, and the reason is a boundary rather than caution:
 INSTALL_UPDATE_HELPER=1 SITE_HOST=chat.example ./deploy/install.sh
 ```
-That installs `lembas-update.path` and `lembas-update.service`. The web
+That installs `lembas-update.path` and `lembas-update.service`, and puts a
 **root-owned copy** of `update.sh` at `/usr/local/lib/lembas/update.sh`. The web
 interface writes `$PREFIX/data/update-requested`; the path unit notices and the
-service runs `update.sh` **as root**, on the configured channel.
+service runs that copy **as root**, on the configured channel.
 **Why a copy.** The unit used to point inside the checkout, and `install.sh`
 clones the checkout *as the service user* — so root was executing a file the
 unprivileged account could rewrite, and one that every update replaces with
 whatever the branch contained. Either turns a compromise of the web application
 into root, and the second needs no compromise at all. The cost is that changing
 `update.sh` needs the installer re-run; the script tells you when its copy has
 fallen behind, and says so loudly if it finds itself running from inside the
 checkout.
 **If you installed the helper before 1.0.0, re-run the installer.** The old
 wiring stays until you do, and the update button cannot fix it — the button runs
 the old unit.
 **What that grants.** Anybody who can administer this web interface can then
 deploy whatever is on the configured branch and restart the service. That is the
 point of it, and it is why it is not the default.
 **What it deliberately does not grant.** The request file carries nothing that
-reaches a command line — no ref, no branch, no channel, no arguments. Both are
+reaches a command line — no ref, no branch, no channel, no arguments, and its
-baked into the unit at install time, so the button is always "deploy the channel
+*contents* are never read at all. Both are baked into the unit at install time,
-this host was configured with" and never "deploy something else". Re-running the installer without the flag removes both units and the
+so the button is always "deploy the channel this host was configured with" and
-marker, and the page goes back to printing the manual command.
+never "deploy something else". Re-running the installer without the flag removes
 both units, the marker and the root-owned copy, and the page goes back to
 printing the manual command.
 A re-run **keeps the channel this host already follows** rather than resetting it
 to `stable`: the channel is declared in `lembas.env` and in the unit, a re-run
 keeps the first while rewriting the second, and an installer that silently moved
 one half was causing exactly the mismatch the Updates page detects.
 Without the helper the page says so and shows `sudo …/deploy/update.sh`, which is
 the same honest degradation the SSH and search extras have.
@@ -1,416 +0,0 @@
 # The 0.9 audit
 A working document, written during the pass over everything that landed between
 `0.8.1` and `0.9.8` -- 16 commits, 163 files, +16,795 / −1,608. It exists so the
 findings survive the session that produced them, and so the security and testing
 stages have something to work from rather than re-deriving it.
 **This file is temporary.** Everything in it either becomes a `CHANGELOG.md`
 line, a paragraph in `CLAUDE.md`, or an entry under "After 1.0.0" in `PLAN.md`.
 It is deleted before the release.
 Each finding carries a disposition:
 - **fixed** -- landed, with a test
 - **later** -- real, but the fix changes what a feature does, so it is the
  user's call and it is written up rather than done
 - **stated** -- correct as it stands, and recorded so the next audit does not
  spend an hour rediscovering that it is deliberate
 ---
 ## The harness and the instruction prompts
 ### {{timezone}} shipped as a hole — *fixed*
 `clock.name_for` returns `""` for anybody who has never chosen a zone, which is
 the default state of every account. `substitute` drops a line only when the
 whole line is **blank** after expansion, and this variable sits inside a
 sentence, so every such request carried:
 ```
 - Times the person gives you are in  unless they say otherwise.
 ```
 Both the comment at `harness.py:191` and the fragment's own hint asserted the
 line disappeared. Neither had ever been true. The sibling call site
 (`schedule/compile.py:102`) has always had the fallback.
 Worth noting *why* the fallback is right rather than merely convenient: `stamp`
 is `clock.now_for(user)`, which already falls back to the server's zone, and
 `{{now}}` already prints its offset. The times were already in that zone.
 Withholding the label from a value the model had been given was not restraint.
 ### The admin preview could not show most of what it previews — *fixed*
 `admin_prompts.py` built its variables with `context_variables(db, user, [], None)`
 — no tools, no chat — and backfilled six names. Every other `requires` gate
 stayed empty, so **eleven fragments could never appear in a preview whatever an
 administrator ticked**: the whole agent surface, both scheduling fragments and
 the helper warning. Editing `tool.agent` and pressing preview showed a system
 message with `tool.agent` missing from it, silently.
 Fixed with sample values, which is what `SAMPLE_DOCUMENTS` already did for
 attachments, plus two controls the "Tools offered" row could not reach — "This
 chat is" (ordinary / scheduled task / helper) and an agent-mode select. The
 samples are gated **exactly as `context_variables` gates the real values**,
 because a preview that admits a fragment the real request would not is worse
 than one that omits it.
 ### Fragments named tools that had been withdrawn — *fixed for Plan mode, later for the rest*
 `resolve_tools` filters per **tool**; `harness._families` re-derives gates per
 **family**. So a family survives on its readers while its writers are gone, and
 the family-gated fragment still names them — under `core.tool_list` saying
 "Anything not named there does not exist here — calling it costs a round and
 returns nothing."
 **Fixed:** Plan mode. `agent/tools.py:1185` withdraws `plan_update` there, and
 `tool.plan_update` was gated on `{{plan}}`, which is set whenever a plan exists
 in any mode. A new `plan_editable` variable — the plan text, blank in Plan mode
 — now gates it. Writing the test found a second instance the audit had missed:
 `context.plan` also said "Change it with plan_update", so that sentence moved to
 the fragment whose job it is.
 **Later:** a read-only helper keeps seven fragments naming fifteen withdrawn
 write tools (`notes_create`, `report_write`, `file_write`, `skill_create`, the
 three schedule writers…). The principled fix is the split `tool.skills` /
 `tool.skills_write` already demonstrates, applied to `notes`, `report`,
 `schedule` and `agent_edits` — four more fragment pairs. That is a prompt
 restructure rather than a bug fix and it is proposed for after 1.0.0. The cost
 today is bounded: `{{tool_names}}` is authoritative and the model has it, so a
 helper wastes at most one round discovering a writer is absent.
 ### Three tool descriptions disagreed with their runners — *all fixed*
 - **`subagent_run` opened "a second model with the same tools"** and contradicted
  itself two sentences later. The child is forced to `ask:False`,
  `subagent:False`, `write:False` with commands pinned to `SAFE_COMMANDS`. The
  leading clause is what a model reads when sizing a task.
 - **`knowledge_get` returned `extracted_text` whole** while every sibling reader
  capped and said so — `fetch` at 20k, `file_read`, the memories block, the
  skill index, the project listing. `MAX_EXTRACTED_CHARS` defaults to 120,000,
  so one call on a long PDF filled an ordinary window with nothing reporting it.
  Its description said "in full", which is why this read as correct: the tool did
  exactly what it claimed. Now `MAX_DOCUMENT_CHARS = 40_000`, cut with the model
  told, which is what `fetch` and `file_read` both do.
 - **`_run_ask_user` read `args["question"]`**, singular, against a schema
  declaring `questions` and a list — so its event always carried an empty
  `query`. Harmless only because the path is documented unreachable, which is
  exactly why nothing caught it.
 ### Fragment text that was false — *fixed*
 `tool.subagent_agent` said a helper "reads and reports … and nothing else, in
 every mode" beside a `write` parameter that makes one write files, and named
 **seven** of the **twenty-three** commands in `SAFE_COMMANDS` — so a model
 avoided commands it was allowed, which costs nothing visible and is therefore
 never reported. Both halves are now pinned by tests against the real list and
 the real schema, because prose and a constant drift the moment one is edited
 alone.
 The distinction the wording now has to carry, and did not before: **commands**
 are fixed to the read-only list in every mode, since an unattended chat cannot
 approve anything; **files** may be written, but only by a helper asked for with
 `write` and only from Edit or Auto.
 ### `tool.background` promises a notification that can be off — *later*
 "When a background job finishes you are told in a new turn" has no `requires`
 for `agents.background_notify`, and the runner itself branches on that flag and
 says "Check on it with `job_output`" instead. One fragment, two behaviours. The
 fix is a variable and a gate, the same shape as `plan_editable`; deferred only
 because it wants the split above rather than a third one-off.
 ### `FAMILY_ASK` has no fragment at all — *later*
 Every other family in `FAMILIES` contributes one. `ask_user` is offered in
 nearly every chat and **all** of its guidance lives in its schema description,
 which is the one thing an administrator cannot edit. Proposed rather than done:
 it is new prompt text, and new prompt text in an audit stage is how an audit
 turns into a feature.
 ### Smaller, all *stated*
 - **`model_name` is resolved on every request and rendered by no shipped
  fragment.** It is documented in `VARIABLES` and available to an
  administrator's override, so it is an escape hatch rather than dead work. The
  cost is one small `select` on a table with tens of rows.
 - **Four pairs of fragments share an `order`**, which undercuts "sparse so a
  custom tool can be slotted between two of these". Ties break alphabetically
  and deterministically. One has a readable consequence: `context.schedule`
  sorts below `context.memories`, so a scheduled task's own instruction appears
  under the reader's remembered facts.
 - **`core.rounds` is dark on a stock install**, since `DEFAULT_CHAT_ROUNDS = 0`.
  Deliberate — `core.keep_working` is its complement and exactly one ever fires.
 - **A row-backed fragment with empty `guidance` can never be selected**, so for
  those "never configured" and "deliberately switched off" are the same state —
  the one distinction the absent-vs-empty convention exists to preserve.
 ---
 ## Controls
 The four sweeps `CLAUDE.md` is shaped around were run by hand over every
 template and **all four came back clean**: 68 htmx verbs against 179 registered
 routes with zero method mismatches, 88 plain form actions and 30
 `fetch`/`htmx.ajax` call sites with zero, every verb-carrying element declaring
 its own `hx-target` or `hx-swap="none"` checked against every ancestor, no empty
 verb attributes, both `form="…"` sites correct, and every `hx-target`,
 `hx-include` and `from:` selector resolving to an id that exists.
 Making each one a **test that sweeps** is Stage 2's work. A one-off audit that
 finds nothing is worth much less than the same audit made permanent.
 ## Capabilities that are built, documented and unreachable — *Stage 2*
 The `Model.params_json` and `library.share` family: the code is right, the route
 validates, and nothing can reach it.
 - **Folder nesting.** `PATCH /api/folders/{id}` handles `parent_id` with a cycle
  guard and `MAX_DEPTH = 8`, `partials/_folder.html` recurses, and `README.md`
  advertises "arbitrarily nested". No template or script submits `parent_id`.
 - **Moving a chat into a folder.** `api/chats.py` handles `folder_id` on PATCH;
  only the composer submits it, at creation. `chat/_composer.html` even carries
  a comment reasoning about "the only way into a folder was to make the chat
  elsewhere and move it".
 - **`Connection.extra_headers_json`** is read into every LLM request and written
  by no form, so its documented use — OpenRouter's `HTTP-Referer` — is
  unreachable. *later*: nothing advertises it, so nothing is currently untrue.
 Written and never read: `Chat.compacted_at`, `User.last_login_at`,
 `Schedule.last_fire_at`, `Schedule.compiled_at`. Neither read nor written:
 `general.require_approval` (its comment says "reserved"), and the `back` form
 parameter on two `admin_models` routes.
 **A third, found by asking where generated images go.** `db.delete(chat)`
 cascades to the message and attachment *rows* and leaves every file on disk;
 `files.remove_files_for_chats` was written for exactly that and was called from
 one place, the temporary-chat sweep. The delete button, a schedule's task chat,
 a helper's chat and account deletion all went straight to `db.delete`.
 Underneath it, a second fault that made the first invisible: `files.claim`
 bound `message_id` and never `chat_id`, so anything uploaded on the *new-chat*
 screen kept an empty `chat_id` for the rest of its life. Six readers filter on
 that column, so for those files the harness did not name them among the attached
 documents, the canvas refused to open them, and the one caller the cleanup
 helper had found nothing to remove. Both fixed, and `chat_service.delete_chats`
 is now the only way a chat is deleted.
 *Stated:* `Message.parent_id` and `content_parts_json` are deliberate forward
 -looking columns; `Chat.archived` is already recorded in `PLAN.md` as a column
 nothing surfaces.
 ## Verified by rendering, not by reading — *Stage 2*
 There is no browser extension on this machine, so `/admin/prompts` and
 `/admin/customization` were checked by rendering them through `TestClient`,
 rewriting the asset URLs to `file://`, and driving headless Chromium against the
 result. Worth recording because the first run of that harness **measured an
 unstyled page**: the templates emit absolute `http://testserver/static/…` URLs
 via `url_for`, not `/static/…`, so the replacement matched nothing and no CSS
 loaded. It reported all five tab panels visible at once — a dramatic finding
 that was entirely an artefact. The harness now fails loudly on any unrewritten
 asset URL.
 What it measured once fixed:
 - `/admin/prompts`, tab bar's on-screen position across the five tabs:
  **385 → 642px before** (the page moving under the reader), **195px constant
  after**, with `scrollTop` staying at 0.
 - `/admin/customization`, document height: **5495px before, 2820px after**.
 ## Documentation — *Stage 2*
 - `README.md` is stale by six phases. "Planned: Image generation · OCR ·
  semantic search" — two of the three shipped. Nothing under Features mentions
  schedules, reports, subagents, branding, quotas, web push, updates, Docker or
  LXC.
 - `CLAUDE.md` says "Three topics live in `docs/notes/`" and lists seven.
 - `CLAUDE.md` says 2088 tests; there are 2092 before this stage.
 - `tests/__pycache__/test_zz_{dump,live}*.pyc` are stale bytecode for two files
  that no longer exist.
 ## Security — *Stage 3*
 **Two privilege escalations in the update helper, both root, both fixed.** The
 helper is the one place this application deliberately crosses a privilege
 boundary, and it crossed it twice more than intended. Neither is reachable
 from the web interface: both need code execution as the `lembas` service
 account first. That is precisely the boundary the unprivileged split exists to
 hold, so "you need a foothold" is the threat model, not a mitigation.
 **1. Root ran a script the service account owns.**
 `ExecStart=/bin/bash __PREFIX__/app/deploy/update.sh` — inside the checkout,
 owned `lembas:lembas`, because `install.sh` clones as that user. So: write your
 payload into `deploy/update.sh`, `touch data/update-requested` (the service
 account owns that directory too), and systemd runs it as root. The web
 interface's `AdminUser` check is not the gate systemd honours.
 There is a second path needing no compromise at all: an update pulls new code
 *as the service user*, and root then executes whatever `deploy/update.sh` that
 pull contained. **Control of the branch was control of root.**
 Fixed by installing a root-owned copy at `/usr/local/lib/lembas/update.sh` and
 pointing the unit there. The cost — improving `update.sh` needs the installer
 re-run — is the right one: root should not execute a script that arrived over
 the network a moment ago. The script warns when its own copy has fallen behind.
 **The old test asserted the vulnerable line**
 (`assert "ExecStart=/bin/bash __PREFIX__/app/deploy/update.sh" in unit`). It
 passed for the life of the feature and pinned the bug in place — the recurring
 failure of this codebase, applied to a privilege boundary: an assertion about
 the text rather than about the property the text was meant to have.
 **2. Root sourced a file the service account can replace.**
 `. "$PREFIX/.deploy-env"`. The file is root-owned, having been written with
 `sudo tee` — but `$PREFIX` is the service account's own directory at mode 755,
 and write permission on a *directory* is all it takes to unlink a file and put
 another there. On the live host `.deploy-env` did not even exist, so it could
 simply be created. `.` runs its contents as root.
 This one survived the first fix entirely, and the helper is what made it
 reachable: before the `.path` unit existed, `update.sh` only ran when an
 administrator invoked it from a shell. Fixed by parsing the two values it wants
 with strict patterns instead of sourcing. The test asserts that **nothing**
 under `$PREFIX` is sourced, rather than naming `.deploy-env`, because the next
 file read from there would have the same problem.
 **Upgrade note:** a host that installed the helper before this keeps the old
 unit, and only re-running the installer moves it. `update.sh` now detects that
 it is running from inside the checkout and says so loudly — otherwise the
 vulnerable hosts are exactly the ones that never hear about it.
 **Also fixed, sub-threshold as a vulnerability but a real bug:** the share
 panel built its `hx-vals` by pasting the search term into a JSON string. Jinja
 escapes the quote for HTML and the parser decodes it again before htmx parses
 the JSON, so a `"` in a search term ended the string and silently stopped every
 checkbox in the panel from submitting anything. `q` was the last key, so an
 injected one would also have won a duplicate-key parse. Built with `| tojson`
 over the whole object now.
 **3. A read-only command that was not read-only.** `SAFE_COMMANDS` — the list a
 **subagent** is pinned to, in every mode, unattended, with no approval card
 possible — contained `find *`. GNU `find` writes files (`-fprintf`), runs
 programs (`-exec … +`) and deletes them (`-delete`), and none of those needs a
 character `policy._UNSAFE` refuses. `rg --pre` is the same shape.
 So the chain was: a parent in **Plan** mode — which promises "reads freely,
 changes nothing" — spawns a helper on a `RISK_READ` tool with no card; the
 helper's `shell_run` survives because `writes_off` drops `RISK_WRITE` and
 deliberately keeps `RISK_EXECUTE`; `find . -maxdepth 0 -fprintf
 ~/.ssh/authorized_keys 'ssh-ed25519 …'` matches `find *` and runs. Prompt
 injection from a page the model just read is enough to start it.
 Fixed with `policy._ACTION`, refusing those flags in `subject()` rather than
 trimming the allow list — a pattern cannot express "and no dangerous flags",
 and "this one looks read-only" is exactly the reasoning that put `find *` there.
 It costs a false refusal on `grep -- -delete`, which is the right direction to
 be wrong in: a refusal asks, an allow does not.
 **4. `0.0.0.0` walked past the loopback guard.** `_literal` answered from
 `is_loopback`, and `0.0.0.0`/`::` are `is_unspecified` — so it returned a
 *decided* `False`, which short-circuited `resolves_here` and skipped the DNS
 half too. `connect()` to either goes to loopback on Linux, so an SSH profile
 pointed at `0.0.0.0` reached this host's own sshd: the one spelling of "this
 machine" that walked past the guard whose whole job is that sentence.
 **5. Push endpoints skipped the SSRF guard.** `POST /api/push/subscribe`
 checked `startswith("https://")` and nothing else, and `send_one` POSTed to it
 with no `check_url` — the only outbound client in the codebase not going
 through the guard. Delivery is triggered by the caller: send a message, close
 the tab, and `_persist` announces it because nobody is following. Checked now
 at subscribe **and** again before the POST, since the row outlives the first
 check.
 **6. A chat could be put in somebody else's folder.** `effective_system_prompt`
 walks up from the chat through its folder and that folder's parents, so this
 reads another account's system prompt through a field that looks like a tag.
 Both paths had it, and `_new_chat`'s is the instructive one: it resolved the
 folder, discarded it when it was not the caller's, and then stored the **raw
 id** anyway — so the ownership check governed which *seeds* were applied and
 not where the chat actually went.
 ### Clean
 Checked and found sound: the branding CSS and custom-theme generation (ids and
 colour values both validated on **read**, so a row written by hand still cannot
 emit a malformed rule; served as `text/css` rather than inline, so there is no
 HTML context to escape); the unauthenticated branding asset route (random
 names, traversal guarded twice, magic-number sniffing, SVG excluded); sharing
 authorisation on every route; the request file's contents reaching nothing;
 `updates._git`'s fixed argv; and the container (non-root, no secret baked, no
 docker socket, loopback only).
 ## Carried forward from earlier stages
 Questions raised before the security stage, answered by it:
 - the three hand-rolled redirect loops each re-run `check_url` per hop
  (confirmed); does each also drop the secret when a hop leaves its origin?
 - `admin_tools.py` calls `check_url(server.url, allow_private=True)`
  unconditionally at save — a syntax check, with the per-row flag governing at
  call time. Confirm that reading is right.
 - the subagent restriction is the one gate that fails open unattended.
  `_child_scope` forcing `ask:False`, `subagent:False`, `write:False` and
  pinning `allow` to `SAFE_COMMANDS` is confirmed present; the question is
  whether every path reaches it.
 - `pyproject.toml` pins no upper bounds and `deploy/update.sh` runs
  `pip install -e` on every update, so a breaking upstream release arrives on a
  button press. pip's `only-if-needed` default limits the blast radius, which is
  why this is a note rather than an emergency.
 ## Testing — *Stage 4*
 2140 tests became **2283**, and writing them found four bugs that reading had
 not — three of them by driving the JavaScript under a Node DOM stub, which is
 the practice `CLAUDE.md` sets out and the reason it does.
 - **The terminal dropped every keystroke after a reconnect.** `socket.onclose`
  closed over the module-level variable rather than its own connection, and
  `close()` queues its event: both reconnect paths close the old socket and
  immediately open a new one, so the old one's `close` arrived afterwards and
  nulled the *live* socket. Output kept coming, because `onmessage` is bound to
  the object; every send gates on `socket && readyState === OPEN`. It also said
  "Disconnected" about a shell that had just reconnected, because
  `closedOnPurpose` had been cleared for the new connection.
 - **Two scripts were loaded twice on `/messages`** — once by `base.html` and
  again by the page. Each is an IIFE with its own state, so four keyboard
  shortcuts toggled their panel twice and therefore did nothing, `/help` opened
  two dialogs, and an `@` mention attached its file twice. A sweep now refuses
  any template re-loading what `base.html` already has.
 - **The microphone had no guard while the permission prompt was up.** State only
  left `idle` after `getUserMedia` resolved, so every click in that window
  opened another stream; only the last was stopped.
 - **A shared skill took its name out of your own library.** `create` checked
  uniqueness with `by_name`, which is scoped to what is *visible* — owned or
  shared — against a `(owner_id, name)` constraint. `documents.create_base` next
  door asks it correctly. Found by writing HTTP tests for `api/library.py`.
 Also: `--ink-faint` failed the 4.5:1 contrast minimum in **both** themes (3.85
 in Moria, 3.19 in Shire), measured by computing the ratios in a headless browser
 rather than by eye. Raised in both.
 **Verified rather than assumed:** the suite now runs on Python **3.11, 3.12 and
 3.14**. It had only ever run on 3.14, while `Dockerfile` ships 3.12 and
 `requires-python` claimed 3.11 — the interpreter most people would actually run
 was the one nothing had tested. Both pass.
 One measurement worth keeping honest: the `slow` marker splits 365 tests that
 stand up a real server, shell, PTY or git repository, but the fast subset is
 111s against 140s for everything. The split is for CI, not for a dramatic
 saving.
 ## Still open after the testing stage
 - `api/library.py`: 28 routes, effectively zero HTTP coverage. Highest exposure
  — auth, ownership and permission gates all live at that boundary.
 - No real `sync_schema` upgrade test exists. `conftest.py` runs `create_all`
  then `sync_schema`, so the differ is only ever exercised as a no-op. Hard rule
  4 has no other safety net.
 - `cli.py`: zero tests, and `create-admin` is the documented bootstrap.
 - SSE has three end-to-end tests; `services/sse.py` has none.
 - `composer.js`, `terminal.js` and `audio.js` — 1,188 lines — have no assertion
  of any kind.
 - The suite has only ever run on Python 3.14. The Dockerfile ships 3.12 and
  `requires-python` claims 3.11.