A PDF that needs a signature, a highlight or a page removed currently sends the user out to a desktop application or, worse, to a web service that wants the document uploaded first. Both are poor answers for a browser whose point is that nothing has to leave the machine. This is a full-tab editor that opens a PDF, marks it up, fills its forms and saves a new copy, entirely locally. Two engines, vendored rather than installed, because an add-on ships as a self-contained folder over the signed update channel and nothing runs a package manager on the way: pdf.js reads and renders, pdf-lib writes. They share no state. Everything in between lives in PDF user space — points, origin bottom-left — which is the one coordinate vocabulary both speak, so a mark survives zooming, rotating and reordering with no conversion table and save-time needs to know nothing about how a page happened to be displayed. The page strip is built from pdf.js's PDFPageView components rather than its PDFViewer, which renders pages in the file's own order and cannot hide, reorder or individually rotate one — three of the features here. Text layers are ours and stay attached for every page, drawn or not, because Theseus's find bar is Chromium's findInPage over the live DOM and a torn-down text layer is a page Ctrl+F cannot see. Canvases are virtualised; a letter page at 100% is 3.4 MB of bitmap. Redaction is the part worth being careful about. A black box over text hides nothing — the text stays in the content stream and comes straight out of a copy-paste — so the editor says so in a modal before the tool can be used, and on save rebuilds each redacted page as an image, which genuinely removes it. Pages that were not redacted are untouched. Form widgets and links are kept, since they were never the leak. Saving never writes over the original: every save reloads the source bytes and replays the session onto a fresh copy, so a botched save cannot poison the next one. Out of scope for this first version: editing the text that is already in the document, and writing XFA forms back (pdf-lib cannot, so those are fill-and- print only, and the editor says so on open).
5.8 KiB
Signed add-on updates
An add-on can be updated at runtime, without waiting for the next Theseus
release, if its addon.json declares an updateURL. The client polls
that URL, verifies an Ed25519 signature against a hardcoded set of
operator pubkeys, stages the new copy under
<userData>/addons-updates-staged/<id>-<version>/, and promotes it on the
next launch — reusing the same backup-then-swap logic as
seedBundledAddons, so any local edits the user made survive as
<userData>/addons-backups/<id>-<oldver>-<timestamp>/.
Client boot flow
initAddons()
├── promoteStagedUpdates() # staged copy wins if newer than installed
├── seedBundledAddons() # bundle wins if newer than what's on disk
└── AddonHost.discoverAndActivate()
Then, 30 s after boot, checkAndStageUpdates() fetches every installed
add-on's updateURL, verifies its signed manifest, and lands any newer
signed version in the staged dir for the NEXT launch to promote.
Publishing an update
One-time: generate the operator keypair
node scripts/generate-update-keypair.mjs "$USERPROFILE/.silentmode/ops"
The private key stays there. Copy the printed pubkey hex into
addon-update-pubkeys.js and ship a Theseus release that bakes it in;
until you do, checkAndStageUpdates skips silently and the client makes
no outbound requests.
Every update
Sign from a clean copy of the committed add-on (git archive HEAD …), not
the working tree, so another session's uncommitted edits don't ship:
node scripts/sign-addon-update.mjs \
<clean-copy>/bundled-addons/screenshot \
https://navigate.st/bns/theseus.x/extensions/screenshot \
"$USERPROFILE/.silentmode/ops/addon-update-key.pem" --out out/
Produces:
out/screenshot-<version>.tar.gz— upload to<base-url>/<id>-<version>.tar.gzout/screenshot-<version>.entry.json— prepend to your liveupdates.json
The channel files live on Sia, under s3://bns/theseus/extensions/<id>/,
and are served through the public gateway as
https://navigate.st/bns/theseus.x/extensions/<id>/… — that is what every
bundled add-on's updateURL points at. Nothing is on the VPS filesystem.
Upload with the Sia uploader (single file → directory prefix, not a full
target path):
node ../Argus/src/lib/sia-upload.js out/screenshot-<version>.tar.gz bns/theseus/extensions/screenshot
node ../Argus/src/lib/sia-upload.js updates.json bns/theseus/extensions/screenshot
Then confirm both URLs answer 200 through the gateway before relying on
the entry. Live channels: screenshot, aegis and pdf-editor.
A first release has no updates.json to prepend to — create one containing
just the new entry and upload both files. Until it exists the client logs a
404 on every check, which is harmless but noisy:
echo '{"addons":[]}' > updates.json # then merge the .entry.json into it
pdf-editor ships about 4 MB of vendored pdf.js and pdf-lib, so its tarball
is far larger than the others'. Nothing in the pipeline cares, but the first
check after a release takes a moment longer.
updates.json schema
{
"addons": [
{
"version": "0.3.0",
"url": "https://addons.silentmode.st/screenshot/screenshot-0.3.0.tar.gz",
"sha256": "hex",
"sig": "base64"
},
{ "version": "0.2.1", "url": "…", "sha256": "hex", "sig": "base64" }
]
}
Order does not matter — the client picks the highest version greater
than what's installed. Keep old entries around only if you want to
support rollback via a manual downgrade tool; the client never picks
anything older than the installed copy.
Signature domain
message = "silentmode.addon-update-v1|<id>|<version>|<sha256>"
signature = Ed25519_sign(operator_privkey, message)
Domain-separated so the operator key can't be tricked into signing something else (a website login token, a wallet message) that happens to have the right shape.
Rotating the operator key
generate-update-keypair.mjsa new one.- Publish next
updates.jsonentries signed by BOTH old and new key (the client accepts a signature from any pubkey inPUBKEYS_HEX). - Ship a Theseus release that adds the new pubkey.
- After enough time has passed for existing installs to upgrade, ship a follow-up release removing the old pubkey. Stop signing with it.
Threat model
- Compromised operator key = update-tunnel compromise. Attackers can
push a malicious add-on to every install that has the endpoint
reachable. Mitigations under consideration:
- multi-sig: require signatures from N of M operator keys (not
implemented — would extend the
sigfield to an array) - on-BCNR pubkey publication: pin the current pubkey set to a BNS record so rotation is auditable (roadmap)
- multi-sig: require signatures from N of M operator keys (not
implemented — would extend the
- Manifest downgrade — a malicious mirror serves an older signed
entry. The client refuses versions ≤ what's installed, so a rollback
attack cannot land an older signed payload as if it were an update.
A user who wipes
<userData>/addons/and boots against a hostile mirror could receive a stale-but-signed copy; the bundled fallback from the Theseus release will beat it viaseedBundledAddonsunless the bundle is older still. - Path traversal in tarball —
tarrefuses..entries by default; we do not pass-P. Extracted manifest'sid+versionare re-checked against the signed values before staging. - Empty pubkey list = feature off. No outbound requests, no updates, no attack surface. Safe default for fresh builds.
Files
addon-updater.js— client-side fetch, verify, stage, promote.addon-update-pubkeys.js— hardcoded pubkey array (edit and rebuild to rotate).scripts/generate-update-keypair.mjs— one-time keygen.scripts/sign-addon-update.mjs— package + sign an update.