Workspace v2

Signet workspace v2 separates authored files, durable state, runtime files, rebuildable cache, imports, transcripts, skills, and secrets. It preserves the existing evidence, Dreaming, ontology, and Source lifecycle.

Layout and ownership

<workspace>/
  AGENTS.md  SOUL.md  IDENTITY.md  USER.md  MEMORY.md  agent.yaml  .sigignore  .gitignore
  skills/                         # root discovery path; independent Git repository
  files/                          # import inbox, not a Source
  data/signet.db                 # SQLite authority
  data/imports/                  # retained managed import originals
  data/snapshots/                # verified snapshots
  transcripts/<harness>/transcript.jsonl
  runtime/                        # logs, locks, temporary files
  cache/                         # tested rebuildable files only
  .secrets/                       # provider-owned or encrypted secrets

The persisted workspace-layout.json selects v1 or v2. v1 uses memory/memories.db, memory/<harness>/transcripts/, .daemon/, and legacy cache paths. v2 uses the layout above. Custom database, transcript, runtime, cache, inbox, managed-import, secret, skills, data, and workspace paths override defaults. Signet refuses unknown or newer layout versions instead of falling back to legacy paths.

Root Git history, remotes, branches, index, and working state are not rewritten by the layout upgrade. Root skills/ stays independently owned and discoverable by harnesses. Configured external Sources stay in place; Signet does not clone them into files/ or managed import storage.

files/ import inbox

files/ is ingress for manual drops and dashboard uploads. It is not a Source root and does not create a local-files provider. Both paths use one durable admission operation and the existing imported-source lifecycle. New imports retain exact original bytes in resolver-owned managed storage before acceptance, for reindexing, export, provenance verification, and recovery. Older imports without retained originals remain valid but are non-reindexable from original bytes. Setup and the layout upgrade never auto-ingest files that were already in the inbox.

Transcripts

JSONL is the canonical transcript file representation. Normal capture does not create Markdown transcript copies. Markdown is available only through explicit export or view generation. Signet retains the indexed session_transcripts representation while episodic evidence, Dreaming, recall, or recovery use it. A mismatch is an explicit diagnostic state; richer evidence is never silently normalized away.

On historical backfill, completed JSONL turns are the authority for the same session. Signet compares ordered roles and content from legacy Markdown and DB rows, retains divergent sources unchanged, logs the disagreement, and withholds the completion marker until reconciled. Identical turns do not duplicate the JSONL record. Live-only turns can still be replaced by a fuller completed snapshot; later disagreements do not silently complete the backfill. Malformed JSONL rows and non-increasing session sequence numbers encountered during comparison also prevent the marker from being written until repaired.

Upgrade a v1 workspace

New workspaces are created on the v2 layout. workspace-layout.json is written when the workspace is created.

An existing v1 workspace is upgraded in place the first time a v2-aware daemon starts on it. The workspace root does not change, and no configured workspace pointer changes. Before the daemon opens the database or binds any workspace path, it takes the daemon instance lock, then renames Signet-owned v1 paths to their v2 locations under the same root:

v1 path	v2 path
memory/memories.db and its -wal, -shm, and -journal files	data/signet.db and matching files
Signet’s transcript.jsonl files in memory/<harness>/transcripts/	transcripts/<harness>/
top-level memory/*--summary.md, *--transcript.md, *--compaction.md, and *--manifest.md	transcripts/
memory/cache/	cache/
memory/imports/	data/imports/
Signet’s own leftovers in memory/: schema backups (memories.db.bak-v* and their sidecars), transcript backfill markers, backups/, and the retired scripts/, tests/, and requirements*.txt templates	data/legacy-memory/
.daemon/	runtime/

Files in memory/ that Signet did not create stay there, including your own notes and harness daily logs such as OpenClaw’s memory/YYYY-MM-DD.md. Harnesses still find them, and Signet’s Git backup still includes them. The daemon removes memory/ only when the upgrade leaves it empty.

Custom path overrides stay authoritative: an overridden component is not moved, and a custom database keeps its -wal, -shm, and -journal files beside it. Inside a custom transcript or data root outside the workspace, the daemon only renames Signet’s own files within that root: <root>/<harness>/transcripts/transcript.jsonl becomes <root>/<harness>/transcript.jsonl, and a custom data root’s memories.db becomes signet.db. Other files there are left alone. The daemon then writes workspace-layout.json with version 2 and preserves the overrides. Root identity files, agent.yaml, skills/, files/, .secrets/, root Git state, and everything else at the root stay where they are. External Sources stay external. Files are renamed, not copied, so the upgrade uses no extra space and the database file is moved intact.

Before the first rename, the daemon records planned moves and each source item’s filesystem identity in .workspace-layout-upgrade.json at the workspace root. .daemon/ moves last. After an interruption, the next start accepts an already-moved item only when its destination matches the recorded identity. The record may only move paths from Signet’s v1 storage into its v2 storage, and the daemon writes layout v2 only after no v1 storage path remains. If the daemon stopped after writing layout v2 but before removing .workspace-layout-upgrade.json, the next start only finishes that cleanup. If an older journal lacks enough identity to verify a completed move, the daemon leaves the workspace on v1 and refuses to start rather than guessing; the journal lists the planned moves for manual recovery.

If the upgrade cannot proceed safely, the daemon reverses any renames it made, leaves the workspace on v1, records the reason in .workspace-layout-upgrade.json, reports it in signet status and GET /api/status (field workspaceLayout.upgrade), and starts normally on v1. This happens when a v2 destination already exists, a new v2 directory is not empty, a path is on a different filesystem than the workspace, a path sits inside a symlinked directory or is itself a relative symlink, a v1 directory cannot be read, or a rename fails. The daemon retries on the next start, so fix the reported cause and restart.

If a launcher recreated .daemon/ after an interruption (for example with startup logs, a pid file, or telemetry), the next start merges its new files into runtime/ only when the runtime directory matches the identity recorded for the completed move. When a name already exists there, the older file is kept beside it with a .before-<timestamp> suffix. Nothing is overwritten. A mismatched runtime directory, or an interrupted record without enough identity to verify it, makes the daemon refuse to start and report the problem. It does not start on a partially moved workspace or guess which copy is current; .workspace-layout-upgrade.json lists the planned renames for manual recovery.

A workspace whose daemon is still running under an older release is not upgraded until that daemon stops, because the new daemon cannot take the instance lock.

Protection and restore status

CLI, API, and dashboard protection surfaces use one component-aware contract. Components include root-authored files, skills, managed originals, SQLite, transcripts, external Sources, runtime, filesystem cache, and secrets. States include protected, missing, stale, degraded, unknown, external, unverified, and excluded-rebuildable. Overall protected requires current protection for required components and valid restore evidence. Git sync alone cannot produce that result.

The restore comparison checks supplied claims for files, SQLite, daemon health, Source identities, transcript roles/provenance/order, recall scope, Dreaming frontier, ontology history/evidence, and harness identity/skills discovery. It does not itself establish independent recovery evidence or issue a valid receipt. The disposable real-daemon fixture currently checks workspace/database health, a persisted SQLite row and integrity, a restored memory through the daemon API, Source ID/generation, and skill discovery. Recall, Dreaming, ontology, and encrypted-secret continuity still require independent recovery probes; until those exist, restore status remains unverified and no valid receipt is published. Restore evidence is bounded; it does not make an older release safe to run against an upgraded workspace.

Upgrades and downgrades

Install the new release and start the daemon; the layout upgrade runs on its own. Check signet status afterward. If workspaceLayout.upgrade reports a blocked upgrade, fix the reported cause and restart the daemon.

After the upgrade, do not run an older release that predates layout support against the workspace. Older binaries that do not understand the persisted layout version are unsupported.