Aplexica User Guide

This guide explains how the Aplexica command line, tray app, and local web UI work together. It is written for the open source Aplexica daemon and avoids private-business, service-specific, or personal information.

For the short install walkthrough, see quickstart.md. For adapter-specific behavior, see adapters/.

What Aplexica Runs

Aplexica installs three user-facing pieces:

• aplexica: the CLI. It installs and controls the background daemon, opens the local web UI, manages artifacts, resolves conflicts, configures routing rules, and performs backup/restore tasks.
• aplexicatray: the desktop tray or menu-bar companion. It watches daemon status and exposes common actions without opening a terminal.
• Local web UI: an embedded browser interface served by the daemon on loopback HTTP. It is opened from the tray or with aplexica web open.

The background daemon watches supported AI coding agents, imports their native files into the Aplexica Canonical Format, and fans compatible artifacts back out to other installed agents. The canonical store lives under the user’s Aplexica data directory and remains local unless the user explicitly configures an optional remote transport plugin.

First Run

The normal setup path is:

aplexica setup --install

The interactive wizard asks whether to enable the tray, enable the local web UI, install and start Aplexica now, and open the web UI. A non-interactive install can accept defaults:

aplexica setup --yes --install

After setup:

aplexica status
aplexica web open

status confirms daemon liveness, agent detection, artifact counts, and active conflict count. web open mints a one-time local bootstrap URL and opens the browser.

CLI Overview

Run aplexica --help for the complete command list and aplexica <command> --help for command-specific flags. The table below explains the user-facing purpose of each command group.

Command	What it does
setup	First-run wizard. Writes local defaults, optionally installs the daemon service and tray autostart, starts the daemon, and opens the web UI.
status	Shows daemon status, installed agents, artifact counts, unresolved conflicts, and optional JSON/watch output.
daemon	Starts, stops, restarts, reloads, installs, uninstalls, and tails logs for the background sync daemon.
web	Opens the local web UI, prints the listener port, issues one-time bootstrap URLs, revokes web sessions, and enables or disables the listener.
tray	Installs or uninstalls the per-user tray autostart entry. The daemon is managed separately.
config	Shows, edits, validates, diffs, and documents layered configuration.
adapters	Lists, enables, disables, and checks supported adapter installations.
sync	Chooses which agents receive synced data (enable, disable, agents), pauses and resumes sync, runs a one-shot reconciliation, and — when given a directory — runs a foreground sync session. The daemon is preferred for day-to-day use.
backfill	Materializes conversation history into this device’s agents past the recent-history cap. Local only; prints a plan by default.
watch	Watches a directory and imports settled files through a named adapter.
list	Lists artifacts in the canonical store.
show	Shows metadata, event history, and payload summary for one artifact.
import	Imports a native agent file or database into the canonical store.
export	Exports a supported artifact back to an agent-native file tree.
convert	Transcodes an ACF bundle into a target adapter’s native format.
verify	Validates a bundle against the schema and re-hashes every event.
backup	Exports the canonical store to a portable .tar.gz bundle.
restore	Restores canonical-store artifacts from a bundle.
backups	Lists native agent-state snapshots. These are snapshots of agent-owned files, not canonical-store bundles.
restore-native	Destructively restores agent-owned files from a native snapshot. A pre-restore snapshot is taken first so the operation is reversible.
snapshot	Adds snapshot events to bound replay cost for an artifact chain.
retention / gc	Runs retention and cleanup passes for snapshots, pruned history, and blob storage.
pin / unpin	Protects or unprotects an artifact from retention pruning and eviction.
delete	Permanently deletes an artifact from the canonical store.
project	Manages the local project registry used to map canonical project IDs to local paths.
pending	Lists project-scoped artifacts whose project is not linked locally yet.
rules	Lists, adds, removes, edits, tests, and reapplies selective-sync routing rules.
tag	Adds, removes, and lists artifact tags.
event	Performs event-level inspection and tag operations.
conflicts	Lists, inspects, and resolves conflicts detected by the daemon.
branch	Lists, creates, renames, archives, unarchives, and deletes artifact branches.
fork	Forks an artifact at a chosen event into a new branch, optionally targeting an agent.
checkout	Sets an agent’s materialization pointer to a selected branch.
log	Prints the event log for an artifact.
diff	Diffs two branches or two events of an artifact.
merge	Merges one branch into another.
resolve	Starts an interactive conflict resolver.
tool	Lists tool artifacts, previews metadata, checks capabilities, and toggles secret sync for a tool.
secret / list-secrets	Manages secret values referenced by tool artifacts without printing secret values by default.
remote	Configures an optional remote-transport plugin. The daemon treats all compatible plugins through the same protocol.
repair	Offline canonical-store repair. repair conversation collapses echo-duplicated turns; repair materialization inspects or drains the native-write retry queue.
doctor	Produces a redacted diagnostic report suitable for support or bug reports.
update	Checks for, or installs, an authenticated update over the signed release channel.
version	Prints build version, commit, and date.
completion	Generates shell completion scripts.
keygen / keygen-age	Generates signing or encryption keypairs for bundle workflows.
hermes	Runs Hermes-specific import/export helpers.
orphans	Inspects and cleans orphaned artifacts.

Common CLI Workflows

Start or restart the full background service:

aplexica daemon install
aplexica daemon start
aplexica daemon restart

Open and secure the local web UI:

aplexica web open
aplexica web issue-token
aplexica web revoke-sessions
aplexica web disable
aplexica web enable

Inspect sync state:

aplexica status
aplexica status --watch
aplexica list
aplexica show <artifact-id>
aplexica event --help

Choose which agents receive data. Aplexica discovers installed agents and imports from them automatically, but it writes to none of them until you say so — the “discover, then await config” default:

aplexica sync agents
aplexica sync enable codex
aplexica sync enable --all
aplexica daemon reload

Pause and resume without disabling anything:

aplexica sync pause --for 2h
aplexica sync pause --agent codex
aplexica sync pause-status
aplexica sync resume

Resolve divergence:

aplexica conflicts list
aplexica conflicts show <artifact-id>
aplexica resolve <artifact-id>

Work with conversation branches:

aplexica branch list <artifact-id>
aplexica fork <artifact-id> --to-agent codex
aplexica checkout <artifact-id> --branch <branch> --agent <agent>
aplexica merge <artifact-id> --from <branch> --into <branch>

Back up and restore:

aplexica backup
aplexica restore <bundle-path>
aplexica backups list
aplexica restore-native --from <snapshot-id> --yes

Giving a Newly Enabled Agent Your History

Enabling an agent does not hand it your whole archive. By default each agent receives only the most recent conversations — ten, unless you have changed sync.convBackfill or a rule’s route.historicalSyncDepth. Everything from that point forward syncs live and is never capped.

To push more history into this device’s agents, use backfill. It prints a plan and writes nothing until you pass --apply:

aplexica backfill
aplexica backfill --agent kilo --apply
aplexica backfill --depth 100 --apply
aplexica backfill --apply

Three things are worth knowing before you run it:

• It is local. The pass writes native session files for agents on this machine and publishes nothing. Cross-device backfill is reserved for a future release behind the sync.cloudBackfill key; --scope cloud is refused today whether or not that key is set.
• A conversation is never materialized into the agent that wrote it. So the per-agent counts differ, and backfilling only an agent’s own conversations plans nothing.
• The counts are eligibility counts, not pending writes. The planner does not check what is already materialized, so an already-backfilled store re-plans the same numbers and the apply pass finds each session current.

A full backfill of a large store creates one native session per conversation per agent. Run the dry run first and read the numbers.

Repairing a Conversation

repair is for the rare case where a conversation’s canonical head is wrong rather than merely out of date. Both subcommands are read-only until you ask otherwise:

aplexica repair materialization
aplexica repair conversation <artifact-id>
aplexica repair conversation --all

repair materialization lists native session writes the daemon could not complete, with the agent and artifact for each. repair conversation collapses turns that a fixed continuation bug echoed back into a thread, and prints the proposed collapse without writing anything.

Applying a conversation repair is a fleet-wide operation, not a per-device one. A peer still holding the duplicates classifies the repaired thread as a stale redelivery and skips it, so a repair applied on one device alone gets undone. Stop every daemon, repair each device, then restart them:

aplexica repair conversation --all --apply --device-id <this-device-id>

Pass --device-id or the repaired head is not eligible for the outbound sweep, and peers keep serving the duplicated copy. Read this device’s id from aplexica status --json (daemonInfo.localDeviceId). The collapse is greedy, so removing one block can expose another; rerun the dry run afterward and repeat until it reports nothing.

Tray App

aplexicatray is the desktop companion for the daemon. It starts at login when installed with aplexica setup --install, aplexica daemon install --tray, or aplexica tray install.

Tray States

State	Meaning
Starting	The tray has launched but has not received a status snapshot yet.
Idle	The daemon is reachable and healthy, with no recent activity.
Active	The daemon has seen recent import or fan-out activity.
Conflict	One or more artifacts have divergent heads that need review.
Paused	The daemon has been quiet longer than the configured paused threshold or sync was stopped from the tray.
Error	The daemon is unreachable or status polling failed.

Tray Menu

Menu item	Function
Status	Read-only line showing current state and watched directory when known.
Conflicts (N)	Shows unresolved conflicts. Clicking a conflict opens the conflict CLI view; overflow opens the full conflict list.
Pending projects (N)	Shows project IDs or paths waiting for a local project link. Clicking a row opens the pending-project CLI flow.
Open Aplexica	Mints a fresh one-time web UI bootstrap URL and opens the default browser.
Open watched directory	Reveals the daemon’s watched directory in the OS file manager.
Pause sync / Resume sync	Stops or starts daemon sync from the tray. Quitting the tray does not stop the daemon.
Pause sync for…	Pauses for 1 hour, 4 hours, until restart, or a custom duration. Timed pauses auto-resume.
Open logs	Opens the daemon log directory.
Open status report	Runs aplexica status in a terminal.
Open config	Opens the local daemon configuration file in the default editor.
About Aplexica	Shows tray and daemon version information.
Quit Aplexica Tray	Closes the tray indicator only. The daemon keeps running.

Local Web UI

The local web UI is served by the daemon on loopback. It is protected by a one-time bootstrap token and session cookies. Open it from the tray with Open Aplexica or from a terminal:

aplexica web open

The sidebar is grouped into Overview, Sync, and System areas. The top bar contains search placeholder controls, the sidebar toggle, and log out. The footer links to the portal source commit.

Dashboard

The Dashboard is the operational home page.

Daemon panel fields:

Field or control	Meaning
Version	Version reported by the running daemon.
PID	Operating-system process ID.
State	Idle, active, or paused state derived from daemon status.
Uptime	How long the daemon process has been running.
Pending imports	Debounced filesystem changes waiting to be imported.
Pause / Resume	Pauses or resumes daemon sync.

Analytics panels:

Panel	Meaning
Agent sync activity	Stacked bars for recent sync events by agent and artifact kind.
Sync in / out by agent	Inbound and outbound event counts per agent.

Other sections:

Section	Meaning
Agents	Cards for installed agents, with status and artifact counts.
Recent activity	Live event rows. Each row shows event type, title, and relative time.

Agents

The Agents page shows supported AI coding agents and whether they were detected on this machine.

Header controls:

Field or control	Meaning
Connected count	Installed agents divided by total supported agents.
Enable sync for all / Disable sync for all	Toggles cross-agent fan-out to all installed agents. Discovery and import still run.

Agent row fields:

Field or control	Meaning
Agent	Human-readable agent name and description.
Status	Active, idle, paused, attention, or not installed.
Artifacts	Count of canonical-store artifacts attributed to that agent.
Adapter	Aplexica adapter version, not necessarily the agent’s own version.
Last activity	Most recent known activity for that agent.
Sync on / Sync off	Per-agent fan-out toggle.

Agent detail fields:

Field or control	Meaning
Status badges	Installed state, sync state, artifact count, and last activity.
Adapter version	Version of the Aplexica adapter.
Agent version	Placeholder for the agent’s native version when detection is available.
Watched locations	Native roots and registered project paths watched for this agent. Registered project paths can be removed.
Namespaces	Namespaces currently associated with the agent’s artifacts.
Recent events	Event type, detail, and timestamp rows for that agent.

Event Stream

The Event stream page combines live server-sent events with paginated history.

Controls:

Field or control	Meaning
Event kind filter	Filters to daemon, agent, imported, synced, checkpoint, refused, conflict, pending, or rule events.
Agent filter	Filters to agents that appear in the loaded events.
Collapse repeats	Folds repeated live pings or repeated artifact history into fewer rows.
Live status badge	Shows live, reconnecting, or disconnected state for the event stream.
Load older events	Fetches the next page of historical events.

Event row fields:

Field	Meaning
Source badge	Live or historical event indicator.
Event type	Human-readable event kind.
Artifact kind	Memory, skill, tool, conversation, or other when available.
Title	Event summary.
Metadata	Agent, path, artifact ID, size, reason, or target-agent details when available.
Timestamp	Local formatted event time.

Routing Rules

Routing rules decide which artifacts fan out where and which tags or restrictions apply.

Rules list fields:

Field	Meaning
Name	Rule identifier. Opens the detail page.
Match	Human-readable summary of artifact type, tags, source, scope, path, or other matchers.
Targets	Agents that receive matching artifacts. Empty target list means all installed agents.
Effect	Additional behavior, such as local-only routing, excluding secret values, assigning tags, or other advanced effects.

Top-level controls:

Control	Meaning
Add preset	Opens recommended rule bundles and applies selected presets.
Add rule	Opens the rule creation form.

Add/edit rule fields:

Field or control	Meaning
Name	Stable rule name. Names are disabled on edit; rename by creating a new rule and deleting the old one.
Mode	Optional sync mode: unset, live, scheduled, or manual.
Scheduled-sync interval (seconds)	Appears when Mode is scheduled. Defaults to 900 seconds, or 15 minutes.
Artifact types	Memory, skill, tool, and conversation checkboxes. No selection matches every type.
Match tags	Tags an artifact must already carry to match the rule. Free-text chips are allowed.
Target agents	Agents that should receive matching artifacts. No selection means all installed agents. The originating-agent option targets the source agent.
Assign tags	Tags added to matching artifacts.
Sync off device	When enabled, matching artifacts may use configured off-device transport. When disabled, matching artifacts stay local.
Include secrets	Whether secret values referenced by tool artifacts are allowed to sync.

Rule detail also shows advanced matchers such as path, branch name, scope, tool kind, source agent, source device, or skill mode as read-only rows when they exist. The practical form preserves those advanced values when saving.

Conflicts

The Conflicts page lists artifacts with divergent heads.

List fields:

Field	Meaning
Artifact	Readable title when available, plus artifact ID.
Kind	Artifact kind, such as memory, skill, tool, or conversation.
Heads	Number of competing heads.
Actions	Opens the conflict detail page.

Conflict detail fields:

Field or section	Meaning
What changed	Analysis summary, recommendation, auto-resolve note, and highlighted differences.
Head A / Head B	Side-by-side details for the competing heads.
Readable summary	Human-readable summary of a head when analysis exists.
First visible content	Primary visible content extracted from a head when available.
Source agent	Agent that wrote the head.
Event ID	Canonical event identifier.
Content SHA-256	Shortened content hash.
Timestamp	Event timestamp.
Payload preview	Expandable JSON or text preview when manual review is needed.
Accept A / Accept B	Chooses one head as the winner.
Edit manual merge	Opens a textarea to paste merged content. Not shown for conversation conflicts.

Auto-resolvable conflicts show the analysis and no manual action requirement.

Forking

The Forking page manages conversation branches.

Conversation search and rows:

Field or control	Meaning
Search conversations	Filters recent conversations by text.
Conversation title	Readable conversation title.
Source agent	Agent that originated the conversation when known.
Description	Additional summary text when available.
Artifact ID	Canonical conversation artifact ID.
Turns	Number of conversation turns.
Branches	Number of branches on the artifact.
Updated	Last update timestamp.

Branch panel controls:

Field or control	Meaning
Fork from	Source branch head or event used as the fork point.
Agent	Target agent for the forked branch.
Branch	Optional branch name. If blank, Aplexica chooses a branch name.
Rationale	Optional note explaining why the fork was created.
Fork conversation	Creates the branch and materializes it for the selected agent.
Checkout branch	Branch to materialize into an agent.
Checkout agent	Agent that should receive the selected branch.
Checkout	Moves the selected agent’s materialization pointer to the branch.

Branch table fields:

Field	Meaning
Branch	Branch name and fork source hash when known.
Events	Number of events on that branch.
Last event	Last event timestamp.
State	Active, archived, or merged.
Materialized	Agents currently materialized on the branch.

Projects

Projects map local folders to project-scoped artifacts.

Add folder fields:

Field or control	Meaning
Path	Absolute or local folder path to register.
Scope	Local for folder-specific memory, global for broader memory scope.
Add project	Saves the project registration.

Project table fields:

Field or control	Meaning
Path	Display name and absolute path.
Scope	Local or global.
Agents	Agents allowed for this project. Empty means all agents.
View memory	Expands effective memory files for the project.
Edit	Edits scope and agent selection.
Remove	Removes the registry entry; it does not delete project files.

Edit fields:

Field	Meaning
Scope	Local or global.
Agents	Installed-agent checkboxes. Leaving all unselected means all agents.

Effective memory fields:

Field	Meaning
File name	Memory file name.
Synced agents	Agents receiving that memory file.
Content	Composed memory content as synced by the daemon.

Pending Projects

Pending projects are folders or project IDs Aplexica noticed but has not fully registered on this device.

Suggestion rows:

Field or control	Meaning
Suggested agents	Agents newly observed in an already registered project.
Add agents	Adds those agents to the registered project’s agent set.
Dismiss	Dismisses the suggestion.

Pending table fields:

Field	Meaning
Path or project ID	Local sample path if known, otherwise canonical project ID.
Agents	Agents associated with the pending row.
Last active	Relative time of the most recent activity.
Type	Discovered folder, artifact-sourced project, or pathless artifact.
Actions	Approve/link or deny.

Approve dialog fields:

Field	Meaning
Path	Folder path being approved.
Scope	Local or global.
Agents	Agent checkboxes. Empty means all installed agents.

Link dialog fields:

Field	Meaning
Local path	Path that should be associated with a pathless or artifact-sourced project ID.

Denied rows can be approved later or restored to the active pending list.

Remote Connection

Some builds show an optional remote connection page when a compatible remote transport plugin is configured. The open source daemon treats this as a plugin protocol and does not require any specific provider.

States:

State	Meaning
Not configured	No remote plugin is configured for this daemon.
Pairing wizard	A plugin is configured but this device is not paired.
Connected card	This device is paired and the plugin reports its connection state.

Pairing fields:

Field or control	Meaning
Pairing code	One-time code or token from the remote provider.
Device name	Optional friendly name for this machine.
Pair this device	Exchanges the code through the plugin and stores device credentials.

Connected fields:

Field or control	Meaning
Device ID	Opaque remote-device identifier.
Account ID	Opaque account or tenant identifier from the provider.
Relay connection	Connected, connecting, starting, disconnected, or unknown.
Test connection	Asks the plugin to verify current connectivity.
Pair a different device / re-pair	Reopens the pairing wizard.
Unpair this device	Removes local remote credentials and stops remote sync on this machine.

Backups

The Backups page manages native agent-state snapshots. These are separate from canonical-store bundle backups.

Metric cards:

Field	Meaning
Protected	Agents with usable native snapshots divided by detected agents.
Remote backup availability	Whether the configured backup destination is available. If no provider is configured, backups remain local.
History	Local and remote backup counts.
Storage	Total displayed size of known backups.

Safety table fields:

Field or control	Meaning
Agent	Agent name.
Status	Protected, backup required, blocked, overridden, or related safety state.
Roots	Number of native roots or the latest error.
Last backup	Relative time of the newest snapshot for that agent.
Backup	Starts a snapshot for that agent.
Override	Allows sync to proceed despite a backup blocker. Use only after understanding the risk.

Manual backup fields:

Field or control	Meaning
Local destination	Writes the snapshot to local storage.
Remote destination	Enabled only when a compatible provider is configured.
Agent checkboxes	Selected agents to snapshot. Empty means all agents.
Back up selected	Starts a snapshot for selected agents and destination.
Back up all	Starts a snapshot for all agents.
Cancel	Cancels a running backup job.

Schedule fields:

Field or control	Meaning
Enabled	Turns scheduled native snapshots on or off.
Presets	Hourly, 6 hours, 12 hours, daily, or weekly intervals.
Custom minutes	Custom interval in minutes, minimum 15.
Destination	Local or remote provider when available.
Agent checkboxes	Agents covered by the schedule. Empty means all agents.
Save schedule	Persists the schedule.
Next run	Next scheduled run when known.

History filters:

Field	Meaning
From / To	Date range filters.
Agent	Shows backups containing a selected agent.
Type	Manual, scheduled, pre-sync safety snapshot, or pre-restore undo snapshot.
Location	Local or remote.
Reset	Clears filters.

History table fields:

Field or control	Meaning
Created	Snapshot creation time.
Location	Local or remote and encryption status when known.
Type	Manual, scheduled, safety, or undo snapshot.
Agents	Agents recorded in the snapshot.
Size	Display size.
Files	Number of files in the snapshot.
Device	Origin device name when known.
Restore scope	For multi-agent snapshots, choose all agents or one agent before restoring.
Restore	Opens destructive restore confirmation.
Delete	Opens deletion confirmation.

Retention fields:

Field	Meaning
Keep	Per-agent number of native backups to retain. Values are clamped between 1 and 100.
Save retention	Persists retention limits.

Delete confirmation requires typing the displayed phrase. Restore confirmation overwrites live native agent files, but Aplexica takes a pre-restore snapshot first so the operation can be reversed.

Settings

Settings exposes safe-to-edit daemon configuration.

Field or control	Meaning
Open raw config	Copies or displays the local config path.
Log level	Trace, debug, info, warn, or error.
Hermes watch interval	How often Aplexica polls Hermes state.db for new or changed conversation sessions. Use a duration such as 5s, 30s, or 1m.
Save	Writes the configuration patch.

Transport

Transport controls how Aplexica reaches beyond this machine when a remote transport is configured.

Status fields:

Field or control	Meaning
Mode	Current transport mode, such as local-only or bring-your-own relay.
Available transports	Modes supported by the current daemon and plugins.
Switch to local-only	Disables off-machine transport without removing local data.

Bring-your-own relay fields:

Field	Meaning
Broker URL	Relay broker URL. Must be a valid URL.
Client certificate path	Optional path to the mTLS client certificate.
Client key path	Optional path to the mTLS private key.
CA certificate path	Optional path to a certificate authority file.
Namespaces	Optional comma-separated namespace list.
Save BYO relay	Saves the relay configuration when the mode is supported.

Help

The Help page links to user documentation, source code, security reporting, and a reminder that the UI can be reopened from the tray.

Onboarding

The onboarding flow walks through initial local setup.

Step	Meaning
Install daemon	Confirms the browser is connected to a running daemon.
Connect your agents	Waits for supported agents to be installed and detected.
Wait for your first sync	Waits for the first sync event.
Done	Links back to the Dashboard.

Troubleshooting

The web UI says the session expired

The local UI session is tied to the daemon. Reopen the UI from the tray or run:

aplexica web open

The tray says the daemon is unavailable

Check daemon status and logs:

aplexica status
aplexica daemon logs --follow

Restart if needed:

aplexica daemon restart

An agent does not appear

Confirm the agent is installed and that Aplexica can discover its native roots:

aplexica adapters list
aplexica status

The daemon imports from installed agents even if fan-out to a specific agent is disabled.

A folder appears under Pending projects

Open the Pending projects page and approve the folder, or link it from the CLI:

aplexica pending list
aplexica project link <project-id> <local-path>

An agent is missing older conversations

Expected. Enabling an agent back-fills only the most recent conversations into it; everything after that syncs live. Push more history in with aplexica backfill — see Giving a Newly Enabled Agent Your History.

Note that a conversation is never materialized into the agent that authored it, so an agent will not show you its own conversations arriving from elsewhere.

A conversation shows the same turns twice

Check whether the duplicate is in the canonical head or only in one agent’s session file:

aplexica repair conversation <artifact-id>

That prints the proposed collapse without writing. If it reports nothing, the canonical thread is clean and the duplicate is in a native session file that will be rewritten on the next materialization.

If it does report a collapse, apply it on every device with all daemons stopped, as described under Repairing a Conversation. Repairing one device alone does not hold: peers still holding the duplicates treat the repaired thread as a stale redelivery and re-serve their copy.

A conversation stopped receiving updates

Canonical sync and the native session file are separate steps, and it is usually the second one that is stuck. Check the retry queue:

aplexica repair materialization

An entry there means the daemon has the turns but could not write them into that agent’s session. The common cause is a session file that diverged from canonical — Aplexica will not overwrite an agent’s own session, so it defers.

Confirm canonical is actually current before assuming sync is at fault:

aplexica log <artifact-id>

If the turns are present in the log, sync is working and only the native write is outstanding. Remote sync runs on a schedule rather than instantly, so a recently authored turn may simply not have swept yet.

A backup restore looks wrong

Native restores take a pre-restore snapshot before overwriting files. List snapshots and restore the pre-restore snapshot if you need to undo:

aplexica backups list
aplexica restore-native --from <pre-restore-snapshot-id> --yes

Publishing This Guide on aplexica.com

The open source guide should have one canonical source:

Aplexica/Aplexica
└── docs/user-guide.md

The website should render that file into HTML instead of maintaining a second hand-edited copy. Recommended implementation:

1. Treat docs/user-guide.md in the public daemon repository as canonical for open source CLI, tray, and local web UI docs.
2. In the website repository, add a build step that fetches the guide from a pinned Git tag or commit, converts Markdown to HTML, and places it under a route such as /docs/open-source-user-guide/.
3. Store the source commit SHA and rendered timestamp in the page footer so users can tell which daemon release the hosted docs describe.
4. Add a CI check in the website repository that fails if the rendered copy was edited manually instead of regenerated from the canonical source.
5. Keep service-specific docs in the website or service app repository, not in the OSS daemon repository. Link between service-specific docs and this OSS guide only where workflows cross product boundaries.
6. Version hosted documentation by daemon release. For example, /docs/open-source-user-guide/latest/ can point at the latest stable daemon tag, while older pages remain available by version.

This keeps open source documentation public and reusable while preventing private service-specific details from leaking into the public daemon repository.