Changelog

Walrus Memory MCP v0.0.18

Project-level MCP config files such as `.cursor/mcp.json`, `.vscode/mcp.json`, or `.claude/settings.json` populate `env`. Before this release, a truthy `MEMWAL_AUTO_SAVE` in those files could override a stored `autoSave: false` or bypass the consent prompt on a new install. The MCP server and plugin hooks now accept only a disabling override (`MEMWAL_AUTO_SAVE=0`) from `env`. They keep `state` and `enabled` consistent and preserve `pendingConsent: true` until `~/.memwal/settings.json` holds a stored answer. Pending SSE readers now wake as soon as the connection aborts, so idle sessions reconnect without waiting for the HTTP reader to fail.

Improvements

• `MEMWAL_AUTO_SAVE` in `process.env` can no longer turn automatic memory on or clear `pendingConsent`. An MCP client populates `env` from `.cursor/mcp.json`, `.vscode/mcp.json`, or `.claude/settings.json` inside the checkout, so a truthy `MEMWAL_AUTO_SAVE` in `env` could flip automatic saving on over a stored `autoSave: false` (reporting `enabled: true` alongside `state: "off"`) or bypass the consent prompt on an unanswered install. Both the MCP server and the plugin hooks now honour only a disabling override (`MEMWAL_AUTO_SAVE=0`) from `env`, keep `state` and `enabled` consistent, and preserve `pendingConsent: true` until a stored answer exists in `~/.memwal/settings.json`.

Walrus Memory TypeScript SDK v0.1.12

This release fixes how `withMemWal` handles long user turns. It caps the recall query at 16,384 UTF-8 bytes and uses `strategy: "head-tail"` to keep both the leading context and the trailing question or instructions. It caps auto-save `analyze()` input at 65,536 UTF-8 bytes. Truncation never splits a multi-byte UTF-8 codepoint across a boundary. The language model prompt still receives the full user turn. Turns longer than the embedder's 16 KiB limit still inject relevant memories without HTTP 400 errors. Auto-save extracts facts across turns of up to 64 KiB.

Improvements

• `withMemWal` bounds the recall query (preserving both head context and tail question/instructions via `strategy: "head-tail"`) to 16,384 UTF-8 bytes and auto-save `analyze()` input to 65,536 UTF-8 bytes without splitting multi-byte UTF-8 codepoints across boundaries, while keeping the full user turn in the language model prompt. User turns exceeding the embedder's 16 KiB limit still inject relevant memories without HTTP 400 errors, and auto-save extracts facts across turns up to 64 KiB.

Walrus v1.59.0

Storage nodes on the public testnet now attest their blob info snapshot on chain at each epoch boundary, because `blob_info_snapshot.certify` defaults to `true` on testnet. To opt out, set `blob_info_snapshot: {certify: false}` in the node config. The default on mainnet and other networks is unchanged. Certification requires the version-4 system contract. On an older contract, attestations fail and are counted, and nothing else changes. The node creates the `blob_info_snapshot_publication` RocksDB column family automatically on first start. New `blob_info_snapshot_*` metrics report durations, errors, and the latest certified epoch. Peak memory for encoding the snapshot drops from about 5.5 to about 1.7 times the snapshot size. Nodes and the CLI no longer read `WALRUS_GRPC_MIGRATION_LEVEL` and always use Sui gRPC, so you can remove that variable from your environment.

Improvements

• adds the `blob_info_snapshot.certify` config flag (default `false`), a new RocksDB column family `blob_info_snapshot_publication` (created automatically on the first start), and metrics under `blob_info_snapshot_*` for serialize, encode, and certify durations and errors, publications found uncertified, and the epoch of the latest certified snapshot. Enabling `certify` requires the version-4 system contract; on an older contract the node's attestations fail and are counted, and nothing else changes. No operator action is required by default.
• encoding the blob info snapshot at the epoch boundary no longer holds the snapshot's full erasure-coded expansion in memory; peak memory for that step drops from about 5.5 times the snapshot size to about 1.7 times.
• The `WALRUS_GRPC_MIGRATION_LEVEL` environment variable is no longer read; the node always uses Sui gRPC. You can remove it from your environment.
• Nodes configured for the public testnet now attest their blob info snapshot on chain at each epoch boundary by default (`blob_info_snapshot.certify` defaults to `true` on testnet). Set `blob_info_snapshot: {certify: false}` in the node config to opt out. The default on mainnet and other networks is unchanged.
• The `WALRUS_GRPC_MIGRATION_LEVEL` environment variable is no longer read; the client always uses Sui gRPC. You can remove it from your environment.

Walrus Memory MCP v0.0.17

This release fixes duplicate storage charges in `memwal_remember_bulk`. Previously, when the bridge reconnected and replayed a bulk call, the call could register a second paid Walrus blob. The bridge now stamps each call with a stable idempotency key before tracking it. The sidecar sends that key to the relayer, and the relayer answers a replayed key with the jobs it already queued instead of creating new ones. This protection applies only to bridge replays. If your agent retries a bulk call itself, that retry counts as a new call and gets a new key. The plugin launcher in this release still installs MCP server 0.0.16. Plugin users get the fix from v0.0.18.

Improvements

• `memwal_remember_bulk` no longer registers a second paid Walrus blob when the bridge reconnects and replays the call. The bridge stamps each call with a stable idempotency key before tracking it, and the sidecar sends that key to the relayer, which answers a replayed key with the jobs it already queued. A bulk call the agent retries itself is a new call and gets a new key.

Walrus Memory TypeScript SDK v0.1.11

`remember`, `rememberAsync`, and `rememberAndWait` accept `{ dedupe: true }` for opt-in content-hash deduplication per owner, namespace, and text, with no time window. Duplicate writes collapse onto the existing live job and `blob_id`. If the memory was forgotten or deleted, the client writes it fresh. `MemWalMock` mirrors this behavior. `MemWal.create` and `MemWalManual.create` copy a supplied `Uint8Array` key, so `destroy()` leaves your buffer intact. `MemWalManual.destroy()` also drops cached secrets and rejects later calls. `formatUntrustedMemories` and `withMemWal` drop non-finite distances and clamp relevance to [0, 1]. `listNamespaces`, `listMemories`, and the manual vector methods validate inputs before signing. `analyze()` accepts an abort `signal`.

Improvements

• `remember`, `rememberAsync`, and `rememberAndWait` accept `{ dedupe: true }` (`RememberOptions` / `RememberWaitOptions`) for opt-in content-hash deduplication per `(owner, namespace, text)` without a 30-minute time window. When `dedupe: true` is set, the client derives a bucket-less `dedupe:<sha256>` key and sends `dedupe: true` to `POST /api/remember`, collapsing duplicate writes onto the existing live job and `blob_id` while writing fresh if the memory was later forgotten or deleted. `MemWalMock` mirrors the same behavior.
• `formatUntrustedMemories` and `withMemWal` validate that each recalled memory's `distance` is a finite number and clamp similarity relevance (`1 - distance`) to `[0, 1]` before filtering and formatting. Records with `undefined`, `NaN`, or `±Infinity` distances are dropped instead of rendering `"relevance":"NaN"`, and negative or greater-than-one distances are clamped to `1.00` / `0.00`.
• `MemWal.listNamespaces` and `MemWal.listMemories` reject a `limit` that is not a positive integer before signing the query string, with the same `limit must be a positive integer, got <limit>` `TypeError` as `recall` and `recallManual`.
• `analyze()` accepts an optional `signal` in `AnalyzeOptions` and aborts the in-flight `/api/analyze` HTTP request when the caller's `AbortSignal` fires.
• `MemWal.rememberManual`, `MemWal.recallManual`, `MemWalManual.rememberManual`, and `MemWalManual.recallManual` reject an empty or non-finite `vector` (including `NaN`, `±Infinity`, non-number elements, and sparse arrays) before signing, and `rememberManual` rejects a missing or blank `encryptedData` string with `TypeError`.

Walrus Memory Python SDK v0.2.2

`MemWal` and `MemWalSync` add `forget(namespace=None)`, with the alias `forget_namespace`. It wraps `POST /api/forget`, deletes the owner's index rows for the namespace, leaves Walrus blobs in place, and returns a `ForgetResult`. `destroy()` and `close(clear_keys=True)` zero the client's copy of the seed, drop the signing key and cached Seal session credentials, and reject later calls with `MemWalError`. `create()` accepts 32-byte `bytes` or `bytearray` keys and copies caller buffers, so `destroy()` leaves yours intact. An empty or non-JSON 2xx response now raises `MemWalJsonBodyError`, a `MemWalError` subclass, instead of `json.JSONDecodeError`. A response missing a required field raises `MemWalError` instead of `KeyError`. `ask()` and `recall()` reject a blank `question` or `query` and an invalid `limit` or `max_distance` before signing. Cancelling one concurrent caller no longer aborts the others during initialization.

Walrus Memory OpenClaw v0.0.8

Auto-capture keeps state for at most 1,000 recent session scopes and expires scopes after 30 idle minutes, so long-lived gateways stop retaining every past session. A per-session message watermark skips already-captured turns, and `occurredAt` uses each message's creation timestamp. `withTimeout` aborts the HTTP request when `requestTimeoutMs` expires, `withRetry` does not retry timed-out calls, and an in-flight guard prevents overlapping auto-capture requests. `memory_store` no longer reports pending jobs as stored. It polls `waitForRememberJobs` when available, surfaces upload failures, and otherwise reports `"Accepted N fact(s) (not stored yet; job_ids=...)"`. The plugin also detects letter-spaced prompt injection, such as `i g n o r e`, and uses MemWal SDK `0.1.10`.

Walrus Memory MCP v0.0.16

This release updates `@modelcontextprotocol/sdk` to 1.32.1, which includes transport security fixes. Authenticated SSE GET and POST requests now refuse redirects. The message endpoint that the server supplies must stay on the configured relayer origin. Because of this, custom authentication headers and tool arguments cannot be forwarded to a redirected origin. If your deployment relies on redirects to reach the relayer, configure the final relayer URL directly.

Walrus Memory TypeScript SDK v0.1.10

This release tightens input and response validation in the Walrus Memory SDK. `recallManual` on `MemWal` and `MemWalManual` rejects `NaN` or `Infinity` in `scoringWeights` before signing, matching `recall`. Previously, `JSON.stringify` sent these values to the relayer as `null`. `remember`, `rememberBulk`, `analyze`, `embed`, and `recall` also reject text containing only whitespace and zero-width characters. Previously, these calls waited until the client timed out. `MemWalMock` rejects the same input. When a 2xx response has the wrong JSON shape, the SDK raises an error naming the endpoint and the missing array instead of a raw `TypeError`. The SDK does not retry these bulk status errors. A null or non-object status body makes `getRememberStatus` return `not_found` and `waitForRememberJob` throw `MemWalRememberJobNotFoundError`.

Walrus Memory Python SDK v0.2.1

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory Python SDK v0.2.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Sites v2.12.2

This release moves the portal Docker images to the `oven/bun:1.4.3` base image on Debian 13. The previous base image reached end of life, which broke `apt` during the server-portal image build in v2.12.1. If you build or run your own portal, the image now builds again. The site-builder is unchanged from v2.12.1, and there are no Move contract changes. `--dry-run` cost estimation still relies on the Sui JSON-RPC `dev_inspect` API, which public Sui fullnodes no longer serve. Regular `deploy`, `publish`, and `update` commands work as before.

Improvements

• server-portal: Move Docker images to `oven/bun:1.4.3` (Debian 13) so `apt` works again in the image build
• Bump version to 2.12.2

Walrus Sites v2.12.1

This release updates the site-builder dependencies. It now builds against Walrus `testnet-v1.58.1` and Sui `testnet-v1.81.0`, up from `testnet-v1.51.2` and `testnet-v1.74.0`, and uses the Rust 1.96 toolchain. The portal and the Move contracts are unchanged. One known limitation remains. `--dry-run` cost estimation still relies on the Sui JSON-RPC `dev_inspect` API, which public Sui fullnodes no longer serve, so cost estimates might not work against those nodes. Regular `deploy`, `publish`, and `update` commands are unaffected. A gRPC-based estimate is planned for a follow-up release.

Improvements

• site-builder: Bump Sui to `testnet-v1.81.0`, Walrus to `testnet-v1.58.1`, Rust toolchain to 1.96
• site-builder: Initialize typed-store DB metrics once in the integration-test harness to fix a startup race
• Bump version to 2.12.1

Walrus Memory MCP v0.0.15

A project-local `.memwal/credentials.json` is ignored until you approve that exact project path, account, delegate key, and relayer with `memwal-mcp approve-project`. Approvals are stored in `~/.memwal/project-approvals.json`, and any change to the destination requires approving again. Until then, the global credentials are used and a stderr line names the skipped file. `memwal_health` now reports `account=` beside `relayer=`. `MEMWAL_CREDS_DIR` must now be an absolute path outside the current project. Other values are refused with an error, and an empty value means unset. The login write-ahead record moves out of the repository to `~/.memwal/login-pending/`. `memwal_login` with `freshKey: true` now closes the previous listener and names the abandoned public key and address so you can remove them from the dashboard.

Walrus Memory TypeScript SDK v0.1.9

`createAccount`, `addDelegateKey`, `removeDelegateKey`, and `MemWalManual` work again without a `suiClient`. The default is now `SuiGrpcClient` against the public fullnode for `suiNetwork`. `rememberBulk` and `rememberBulkAsync` send a derived idempotency key, so a retry within 30 minutes returns the same jobs. `listMemories()` pages stored memories as metadata without a similarity query. Remember-job waits back off toward a 5s cap. On timeout they raise `MemWalRememberJobTimeout` with `lastStatus` and `serverError`, and they raise `MemWalRateLimited` when status reads stay rate-limited. Many methods reject invalid input before signing, including bad `limit`, `maxDistance`, job ids, and blank text. Package subpaths now resolve in Next.js, Turbopack, and Node `require(esm)`, and `/manual` no longer imports `@mysten/walrus`.

Improvements

• `rememberBulk` and `rememberBulkAsync` send an idempotency key derived from the batch (a 30-minute bucket plus each item's namespace and text). A retry inside that window returns the same jobs instead of storing the batch again. An explicit `idempotencyKey` still wins.
• `withMemWal` does not recall or auto-save on a tool-execution step. A later tool step reuses the recall cached for that user message. Auto-save still runs only on a user turn.
• `flush()` raises when a background auto-save failed, and `onSaveError` runs. The chat turn itself still does not fail.
• `waitForRememberJob` and `waitForRememberJobs` do not sleep past the deadline. If the budget runs out during the sleep, they make one more status read.
• `remember`, `rememberAsync`, and `rememberAndWait` reject a non-string text argument with `TypeError` before signing. An object used to be sent as `text` and the relayer answered 422.
• `MemWalMock.rememberBulkAsync` rejects an empty or non-array batch with the same error as `MemWal`.
• `MemWalMock` honors `idempotencyKey`. A retry of the same text and namespace returns the original job, a different payload throws, and `forget` / `clear` release the key so the text can be written again.
• `MemWalMock.recall({ sort: "recent" })` orders a bounded semantic window (5× `limit`, at least `limit`, at most 50) by write time before truncating. Relevance order is unchanged.
• Positional `recall(query, options, namespace)` uses that namespace when `options.namespace` is omitted, on both `MemWal` and `MemWalMock`. An explicit `options.namespace` still wins.
• `remember(text, { idempotencyKey })` and `rememberAndWait(text, { timeoutMs }, { idempotencyKey })` treat a plain object in the namespace slot as options. A third options argument is merged, and its fields win. That call used to send the object as the namespace, or drop the third argument and mint a second blob. An explicit string namespace still wins.
• `getRememberStatus("")` returns `not_found` without a request, and `waitForRememberJob("")` fails immediately with status 404. `GET /api/remember/` is not a route; the empty 404 body used to throw `SyntaxError`, and the wait treated that as transient until timeout.
• `rememberBulkAsync` rejects more than 20 items before signing, matching the relayer's `MAX_BULK_ITEMS`.
• Remember-job status reads (`getRememberStatus`, `waitForRememberJob`, `getRememberBulkStatus`) no longer attach `x-seal-session`. Status is metadata and does not decrypt.
• Bulk status includes each job's `namespace`. `waitForRememberJobs(jobIds)` uses that value when the caller did not pass one, instead of the client default. A caller-supplied namespace still wins.
• `getPublicKeyHex()` and any request after `destroy()` throw, including a read already in flight. `getPublicKeyHex()` used to resolve to the zeroed key, or to the well-known key of the wiped seed, and `remember()` still sent `GET /version` before noticing the client was destroyed.
• `normalizeServerUrl` no longer warns about plaintext HTTP for IPv6 loopback. `URL.hostname` keeps the brackets on `[::1]`, so the `::1` exemption never matched `http://[::1]:8000`.
• `remember`, `rememberBulk`, `analyze`, and `embed` reject whitespace-only text immediately, with the same messages the relayer returns. A blank body used to be accepted and then sit until the client timed out.
• A 401 `AUTH_REJECTED` now says a wallet address is not a MemWal account ID. The relayer still returns one 401 for every credential failure, so the message names the mix-up instead of revealing which check failed.

Walrus Memory Python SDK v0.1.12

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.58.1

This release adds the `blob_info_snapshot.certify` config flag for storage nodes. The flag defaults to `false`. The storage node also uses a new RocksDB column family, `blob_info_snapshot_publication`, which it creates automatically on the first start after you upgrade. New `blob_info_snapshot_*` metrics are available for monitoring. No operator action is required.

Improvements

• adds the `blob_info_snapshot.certify` config flag (default `false`), a new RocksDB column family `blob_info_snapshot_publication` (created automatically on the first start), and `blob_info_snapshot_*` metrics. No operator action is required.

Walrus Memory OpenClaw v0.0.7

`memory_search` and `memory_store` now operate only within the calling agent's namespace. If you omit the namespace, the tools use that agent's namespace rather than the main namespace. A namespace that the model supplies is rejected unless it matches the agent's namespace or its legacy namespace. Configuration interpolation of `${ENV_VAR}` now applies only to `privateKey`. All other fields, including `serverUrl`, are read literally, so a secret cannot end up in `serverUrl`.

Walrus v1.57.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory MCP v0.0.14

The plugin no longer starts the MCP server through `npx`, which could run a package planted in the open project. Every launch site now runs `plugin/scripts/launch_mcp.mjs`. It installs the pinned version under `~/.memwal/runtime` and fails rather than falling back. `MEMWAL_MCP_RUNTIME_DIR` relocates that directory and must be an absolute path. `memwal_remember` and `memwal_remember_bulk` now return a `job_id` at accept by default. Set `MEMWAL_MCP_REMEMBER_WAIT_MS=90000` to wait for a `blob_id`. Set `MEMWAL_MCP_TRANSPORT=http` to use Streamable HTTP, though replayed writes might duplicate. Timeouts report the cause and relayer health. The bridge honors `retry_after`, `memwal_remember_status` accepts `job_ids`, and delegate keypairs persist before browser sign-in.

Improvements

• The cold-start tool list no longer advertises a tool the relayer may not serve. The bridge ships on npm and updates itself while a relayer ships per environment, so 0.0.14-dev.0 dialled prod and staging still on 0.0.13: cold start named `memwal_remember_status`, which neither registers, and the pending-write wording sent the agent to go call it — one live run spent 90.67s there before erroring. Cold start is now a floor rather than a forecast (`BASELINE_RELAYER_TOOLS`): it carries only what the oldest supported relayer serves and its descriptions name nothing outside it, while newer tools still reach the client a beat later on the relayer's own `tools/list`. Initialize `instructions` use the same floor (they used to name `memwal_remember_status` before any `tools/list`). And a call for a tool outside that floor — including during the cold-start window, before any upstream `tools/list` has been seen — is now answered locally and at once — naming the tools that do exist and saying plainly that nothing ran — instead of being forwarded into a wait that only ends at the orphan deadline.

Walrus Memory TypeScript SDK v0.1.8

All relayer requests now carry a deadline, so one stalled request no longer runs past `timeoutMs`. Requests default to 30 seconds, and you can change this with `requestTimeoutMs` on `MemWal.create`. `restore` and `analyze` each use 60 seconds. An expired request raises `MemWalRequestTimeout` with `status: 504`, which the existing poll handling retries. Generated idempotency keys now derive from the namespace, the text, and a 30-minute bucket. A replay from a new client therefore no longer creates a duplicate paid Walrus blob. Explicit `idempotencyKey` values are unaffected. `recall()` also sends `deadline_ms`. Relayers that read it return a 504 with `serverCode` `RECALL_TIMEOUT` that names the stuck step.

Walrus Memory Python SDK v0.1.11

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.56.1

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.56.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory MCP v0.0.13

Timed-out writes from `memwal_remember`, `memwal_remember_bulk`, and `memwal_analyze` now warn that the write might have completed and tell you to check with `memwal_recall` before re-saving, which avoids duplicate paid copies. The bridge reports a rejected delegate key immediately and points to `memwal_login`. It backs off on HTTP 429 using `Retry-After`. It fails requests stuck behind a failing handshake after 90s, which you can override with `MEMWAL_MCP_STALLED_HANDSHAKE_MS`. After an in-session `memwal_login`, the bridge keeps reading stdin and confirms the sign-in. The CLI writes credentials as a new `0600` file, warns about unrecognized options, and lists network presets in `--help`. `memwal_health` now reports the relayer URL, and `memwal_restore` retries a page after a transient failure.

Improvements

• Unrecognised options now warn on stderr and in the structured log (`cli.unrecognised_arg`) instead of being dropped in silence, so a typo'd `--namesapce work` no longer writes to the default namespace with nothing to say it had. The warning names the option key only, keeping a mistyped value-taking flag (`--tokenn=hunter2`) from putting the secret on stderr, and it warns rather than exits so an option from a newer config cannot brick the server.
• `--help` now lists the network presets (`--prod`, `--dev`, `--staging`, `--local`) with the relayer and web URLs each resolves to, rendered from the preset table rather than retyped so a new preset cannot ship undocumented the way `--prod` did. The `--label` default is corrected to "MCP Client", which is what the code actually falls back to.
• `memwal_health` reports `relayer=<url>`, naming the origin this process dialled, so a client bound to the wrong network finds out there instead of by noticing its memories are missing. The URL is captured when the call is sent rather than when the reply lands, so a reconnect mid-flight cannot label the answer with a relayer it did not come from.
• Keep reading stdin after an in-session `memwal_login`. The auth-required stub hands off to the bridge by pausing stdin, and a stream paused that way does not resume when a new `data` listener attaches, so the bridge served only the request replayed from the hand-off and then went deaf. Signing in appeared to work, the retry succeeded, and every call after it hung until the client timed it out.
• Confirm a completed sign-in instead of only writing it to the log file. The failure path already reported itself twice (a notification and a notice on the next tool call) while success reported nothing, so a user who approved in the browser could not tell whether credentials had landed, the bridge had adopted them, or a retry was worth trying. Success now sends the matching notification and prefixes a one-shot banner naming the account, the delegate, and the resolved credentials path onto the next tool result.
• Stop the `memwal_login` prompt claiming you are already signed in when you are not. The bridge assumed it only ever runs with credentials on disk, but `memwal_logout` deletes them and login is intercepted before the signed-out guard, so a login after a logout in the same session announced that a stored delegate key would be replaced, reading as though the logout had not taken. Both prompts now read the credentials file instead of assuming the mode.
• Write the credentials file by creating a new `0600` file and renaming it into place, instead of writing the delegate private key to the existing path and tightening the permission afterwards. `writeFileSync`'s `mode` only applies when it creates the file, so a `credentials.json` left world-readable by anything outside the CLI (a manual `chmod`, a restored backup, another tool) received the new key under the old permission until the following `chmod` landed. The CLI now writes the backup it takes when signing in as a different account the same way; that backup holds the same plaintext key, and the CLI previously created it under the process umask. On Windows, where `rename` cannot replace a destination another process holds open and POSIX mode bits are not enforced at all, the CLI retries briefly and then writes in place rather than failing the sign-in.

Walrus Memory TypeScript SDK v0.1.7

`restore()` results now include a `failed` count for permanent decrypt or UTF-8 failures. These were previously folded into `skipped` or dropped silently. The field is required, like `truncated`, and the SDK defaults it to `0` when omitted. Request body hashing now uses `@noble/hashes` instead of WebCrypto with a `node:crypto` fallback. Browser apps built with Vite no longer crash on the first remember or recall. The package now declares `engines.node >= 20.0.0`. Empty-body 401 responses return the AUTH_REJECTED troubleshooting message instead of suggesting `memwal_login`. Transactions in `account.ts` and `manual.ts` now use typed `tx.pure` helpers that work with modern `@mysten/sui`.

Walrus Memory Python SDK v0.1.10

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory TypeScript SDK v0.1.6

`recall()` results include an optional `created_at` field with the RFC3339 write-time of the stored fact. `RecallOptions` accepts `sort: "relevance" | "recent"` and `scoringWeights`. With `sort: "recent"`, the SDK over-fetches semantic candidates (5x `limit`, capped at 50), orders them by write-time descending, and then truncates the results to `limit`. The SDK now reports an HTTP 503 with `x-auth-error: AUTH_UPSTREAM_UNAVAILABLE` as a retryable credential-verification outage instead of a sign-in failure. Other 503 responses keep the generic sanitized body, and empty-body 401 responses still point you to `memwal_login`.

Walrus Memory MCP v0.0.12

`memwal_recall` accepts an optional `maxDistance` cosine-distance cutoff, which must be 0 or greater. It drops hits with `distance >= maxDistance`, and result lines now show both `score` and `distance`. The server forwards the client's `initialize.clientInfo` to the relayer as `x-memwal-client` and `x-memwal-client-version`, so sidecar logs name the coding agent. The credential directory now resolves on every access, and `MEMWAL_CREDS_DIR` overrides it. This fixes a Windows problem where the test suite overwrote `~/.memwal/credentials.json` and its delegate key. `memwal_restore` now clarifies that `truncated=true` is retryable and that `truncated=false` does not guarantee completeness.

Improvements

• Optional `maxDistance` on `memwal_recall`: cosine-distance cutoff (low = similar, must be >= 0). Hits with `distance >= maxDistance` are dropped. Result lines now include both `score` (`1 − cosine distance`) and `distance`.
• Resolve the credential directory on every access instead of freezing it at module load, and let `MEMWAL_CREDS_DIR` override it. The login test sandboxed the home directory with `HOME` alone, which `os.homedir()` ignores on Windows, so running the package's test suite there wrote fixture credentials over the developer's real `~/.memwal/credentials.json` and destroyed the delegate key stored in it.

Walrus Memory Python SDK v0.1.9

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.55.2

This release adds one storage node metric, `node_recovery_baseline_invalidations_total`. It counts the sync-and-recovery info publications that invalidate the node-recovery baseline and supersede in-flight node-recovery runs. If you operate a storage node, you can add this counter to your monitoring to track how often in-progress node-recovery runs are superseded. The release body describes no other changes to storage node behavior, configuration, or client-facing APIs. The same change ships in both the Mainnet and Testnet releases of this version.

Improvements

• Adds a `node_recovery_baseline_invalidations_total` metric counting sync-and-recovery info publications that invalidate the node-recovery baseline and supersede in-flight node-recovery runs.

Walrus v1.55.1

This Testnet release adds one metric to the storage node. The `node_recovery_baseline_invalidations_total` metric counts sync-and-recovery info publications that invalidate the node-recovery baseline. Each publication it counts also supersedes any in-flight node-recovery runs. If you operate a storage node, this metric shows how often baseline invalidations supersede in-flight node-recovery runs.

Improvements

• Adds a `node_recovery_baseline_invalidations_total` metric counting sync-and-recovery info publications that invalidate the node-recovery baseline and supersede in-flight node-recovery runs.

Walrus Memory TypeScript SDK v0.1.5

`rememberManual` now sends `encryptedData`, a base64 SEAL ciphertext, instead of a pre-uploaded `blobId`, matching `POST /api/remember/manual`. `recall()` results include `dropped_count` when the relayer omits matches that fail to download or decrypt. `health()` surfaces an optional `write_ready` field when the relayer reports write-path sidecar liveness. Manual recall now decodes SEAL object IDs with `hexToBytes`. This removes the risk of coercing `NaN` to `0` or throwing on empty input. HTTP error messages and remember-job error strings no longer expose `localhost` or `127.0.0.1` URLs to callers.

Walrus Memory Python SDK v0.1.8

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory MCP v0.0.11

This release fixes MCP bridge reliability. The SSE handshake socket no longer aborts on HTTP 429, 401, or a bad content-type, which crashed on Windows. Outbound MCP POSTs are serialized, and the bridge reconnects when the SSE handle is gone. A burst of `memwal_recall` calls no longer breaks it until restart. `login` without a TTY exits with code 1 instead of reporting success. Concurrent `memwal_login` calls reuse the in-flight listener and URL rather than starting a second flow that can hang later remember and recall calls. A login timeout warns that an on-chain delegate key might already exist if the browser approved after the listener expired. Session occupancy caps are reported as concurrent limits rather than a 30-second cooldown.

Walrus Memory TypeScript SDK v0.1.4

This release improves the error message for empty-body 401 responses. If you are calling through MCP for the first time and receive an empty-body 401, the SDK now tells you to run `memwal_login`. Previously it showed `<no message>`, which gave no indication of what to do next. Credential 401s keep their existing behavior and still point you to the AUTH_REJECTED troubleshooting guide.

Walrus Memory MCP v0.0.10

The MCP server now sends proactive-usage instructions in the MCP `initialize` handshake. The model knows when to save and recall memories even when your client lazily loads tool descriptions. The bridge carries these instructions at cold start. Signed-out sessions get guidance pointing at `memwal_login` and a conservative `tools/list`. `memwal_recall` is advertised as read-only. The UserPromptSubmit hook lets the agent choose between remember and recall from meaning rather than keywords. Tool calls whose upstream response never arrives return a retryable error after `MEMWAL_MCP_CALL_TIMEOUT_MS` (default 240s) instead of hanging. `memwal_remember_bulk` uses your configured namespace (`--namespace` / `MEMWAL_NAMESPACE`) when a call omits one. `serverInfo.version` reports the real package version.

Improvements

• Answer orphaned tool calls whose upstream response never arrives with a retryable error instead of hanging indefinitely. The bridge now tracks in-flight request start times and sweeps expired calls through a per-request deadline (`MEMWAL_MCP_CALL_TIMEOUT_MS`, default 240s), reuses the existing late-reply drop so an expired call cannot get a second response, and enriches reconnect logs with pending request IDs and methods.
• Inject the configured default namespace (`--namespace` / `MEMWAL_NAMESPACE`) into `memwal_remember_bulk` calls that omit one, so bulk facts land in the project namespace instead of the relayer fallback `default`.
• Send proactive-usage instructions in the MCP `initialize` handshake, so the model knows when to save and recall without being asked. Clients moved to lazy tool loading, which keeps tool descriptions out of the model's context until a tool is explicitly loaded; the guidance lived only in those descriptions, so the model stopped using memory on its own and would offer its built-in memory or deny the tool existed. `instructions` travels with `initialize`, before any `tools/list`, so lazy loading cannot strip it.

Walrus Memory OpenClaw v0.0.6

`openclaw plugins install` now succeeds, and the plugin declares `contracts.tools`, so `memory_search` and `memory_store` register. Config validation happens at register time with per-field errors and leaves the gateway running. Every relayer call now has a deadline, set with `requestTimeoutMs` (default 10000, range 1000 to 60000). A hung relayer no longer blocks the agent turn. The prompt-injection filter no longer silently drops ordinary statements such as "I run the deploy command every Friday." The docs now cover `hooks.allowConversationAccess`, which auto-capture requires. They also correct the config key: both `plugins.slots.memory` and the `plugins.entries` key use `memory-memwal`.

Walrus Memory TypeScript SDK v0.1.3

The SDK now estimates tokens on the client side and supports token budgets for recall. When recalled content exceeds a budget, you can choose how the SDK truncates it by setting one of the configurable truncation strategies. This release also changes the default SDK relayer URL to `https://relayer.memory.walrus.xyz`. If your application relies on the default relayer rather than setting one explicitly, it uses the new URL after you upgrade.

Walrus v1.54.0

The aggregator now returns 404 `BLOB_NOT_FOUND` within about a second for requests for quilt patches of expired or nonexistent quilts. Previously, these requests returned 503 `BLOB_UNAVAILABLE` after a 10 to 20 second fan-out to storage nodes. In the CLI, `walrus read-quilt` now fails immediately with "the blob ID does not exist" when you read an expired or never-certified quilt. Previously, the command timed out against the storage nodes.

Improvements

• Requests for quilt patches of expired or nonexistent quilts now return 404 `BLOB_NOT_FOUND` within about a second, instead of 503 `BLOB_UNAVAILABLE` after a 10-20 second fan-out.
• `walrus read-quilt` fails immediately with "the blob ID does not exist" for expired or never-certified quilts, instead of timing out against the storage nodes.

Walrus Memory MCP v0.0.8

The Walrus Memory MCP server now includes human-readable titles for its tools during pre-login tool discovery. Each tool also carries explicit `readOnlyHint` and `destructiveHint` metadata at that stage. This metadata matches what the remote relayer provides to Claude connectors, so clients that inspect tools before login get the same titles and hints as they would from the relayer.

Walrus Memory MCP v0.0.7

This release contains three fixes. After you log in through the browser, the MCP server now reloads your credentials, so you no longer need to restart the MCP client. Handshake handling is stricter. The server rejects stale handshakes and prevents requests from being replayed across account changes. Restores now report when results are truncated, and the server enforces the relayer's bounded restore limit.

Walrus Memory Python SDK v0.1.7

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory TypeScript SDK v0.1.2

This release adds deterministic offline mock clients, so you can run tests without credentials. It also adds idempotency keys, which collapse retries onto a single paid remember job. Fixes isolate foreign manual-recall blobs and safely encode large encrypted payloads. The SDK now supports both Zod 3 and Zod 4 as peer dependencies.

Walrus Memory TypeScript SDK v0.1.1

This release adds `flush()`, which you can call to handle pending auto-saves from `withMemWal`. Restore results now include a `truncated` field that indicates when a recovery is incomplete, so your code can detect partial restores. `AUTH_REJECTED` errors now link to troubleshooting guidance, giving you a direct path to resolving authentication rejections.

Walrus Memory Python SDK v0.1.6

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.53.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Memory TypeScript SDK v0.1.0

Recalled AI memory now arrives as nonce-delimited, explicitly untrusted data instead of content injected directly into the system role. This release adds sponsor authorization helpers for sponsored transaction flows. It also adds migration-aware account and manual configuration, including registry support and support for current Sui transaction results. Account transaction result handling and failure diagnostics improve across both legacy and current Sui response shapes.

Walrus Memory MCP v0.0.6

This release tightens the delegate-key login flow. Before it accepts a login callback, the MCP server now requires a localhost preflight handshake that proves the exact state, public key, and relayer. The server also takes the canonical login label from the verified local flow and no longer trusts the label that the browser callback supplies. MCP SSE POST messages and replayed requests now authenticate with the delegate credentials.

Walrus Memory Python SDK v0.1.5

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.52.1

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.52.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Sites v2.12.0

The portal adds a glob-based matcher for routes and redirects, available behind a flag. The matcher enforces a cap on the number of patterns. This is the first step of the portal's routing migration. The site-builder no longer relies on client methods that only work through JSON-RPC, ahead of the JSON-RPC deprecation. This release also updates the site-builder's Walrus dependencies to testnet-v1.51.2.

Improvements

• portal: Glob route/redirect matcher behind a flag, with a pattern cap
• site-builder: Migrate off JSON-RPC-only client methods and bump walrus deps to testnet-v1.51.2
• Read the release tag from the event payload
• Bump version to 2.11.1 (#726), then 2.12.0

Walrus v1.51.2

This release adds a `--commission-receivers` flag to the `walrus info committee` command. When you pass the flag, the CLI prints the commission receiver of every node in the previous, current, and next committees in CSV format. For nodes whose commission receiver is an object, the output also includes the address that owns that object. The paired Testnet release of this version includes the same change.

Improvements

• `walrus info committee` gains a `--commission-receivers` flag that prints, as CSV, the commission receiver of every node in the previous, current, and next committees (including the owning address for object receivers).

Walrus v1.51.1

The `walrus info committee` command now accepts a `--commission-receivers` flag. When you set it, the CLI prints the commission receiver of every node in the previous, current, and next committees in CSV format. For nodes whose commission receiver is an object, the output also includes the address that owns that object. This release includes the same change on Mainnet and Testnet.

Improvements

• `walrus info committee` gains a `--commission-receivers` flag that prints, as CSV, the commission receiver of every node in the previous, current, and next committees (including the owning address for object receivers).

Walrus Sites v2.11.0

The portal now uses gRPC instead of JSON-RPC for its reads from Sui. This change prepares the portal for the deprecation of the Sui JSON-RPC API. If you operate your own portal, it now communicates with Sui over gRPC. The site-builder updates its Sui and Walrus dependencies in this release.

Improvements

• portal: Migrate Sui reads to gRPC
• site-builder: Update Sui and Walrus deps

Walrus v1.50.0

A maintenance release on Testnet with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Package v0.0.5

Fixes a workspace dependency that broke npm installs outside the monorepo and switches the release workflow to pnpm publish.

Package v0.0.5

Adds the automatic memory plugin for Claude Code, Codex, Cursor, and Antigravity, along with new `memwal_remember_bulk` and `memwal_health` tools and proactive memory behavior. Also fixes the plugin bundle so it ships `.mcp.json` correctly.

Package v0.0.4

Fixes HTTPS dashboard sign-in callbacks and credential reload after login so tools work without restarting the MCP client.

Walrus Sites v2.10.0

The portal now bounds each aggregator fetch attempt at 10 seconds and fails over to the next aggregator URL. It sizes Bun's `idleTimeout` from the full retry chain. A slow aggregator no longer drops the inbound connection mid-request and causes a generic upstream proxy error. Instead, the portal returns its own `aggregatorFail()` page. Portal operators can tune this behavior with `AGGREGATOR_REQUEST_TIMEOUT_MS` (default `10000`) and `PORTAL_IDLE_TIMEOUT_MAX_S` (default `100`). The portal sanitizes and logs the resolved values at startup. The site-builder updates Sui to `testnet-v1.73.0` and Walrus to `testnet-v1.50.0`. This release has no Move contract changes.

Improvements

• portal: Size Bun `idleTimeout` from aggregator retry budget; add per-attempt fetch timeout and sanitize-and-log resolved config at startup
• site-builder: Bump Sui to `testnet-v1.73.0`, Walrus to `testnet-v1.50.0`

Package v0.0.3

Rebrands package metadata from MemWal to Walrus Memory throughout the MCP server.

Package v0.0.4

Adds temporal anchoring through `occurredAt` to the `memory_store` tool. The auto-capture hook now passes the current timestamp, and the SDK is bumped for the new `AnalyzeOptions`.

Improvements

• Wire temporal anchoring through the agent-side memory tools.

Package v0.1.4

A patch release for the Walrus Memory Python SDK.

Package v0.0.7

Adds an optional `occurredAt` parameter to `analyze()` for temporal anchoring of extracted facts. Relative date references are now resolved into absolute dates.

Improvements

• Add optional `occurredAt` to `analyze()` and `analyzeAndWait()` for temporal anchoring of extracted facts.

Package v0.0.6

Adds `RecallParams` for object-style recall calls and deprecates the positional recall overload. Also documents `restore()` response fields.

Package v0.1.3

A patch release for the Walrus Memory Python SDK.

Walrus v1.49.1

A maintenance release on Mainnet with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Package v0.0.3

Rebrands from MemWal to Walrus Memory across the OpenClaw memory plugin.

Package v0.0.2

Adds relayer compatibility checks and rebrands from MemWal to Walrus Memory across the MCP server package.

Package v0.0.5

Adds relayer compatibility checks, `RecallOptions` for `topK`/`namespace`/`maxDistance` filtering, and prefers Sui gRPC for SEAL sessions. This release also rebrands from MemWal to Walrus Memory.

Package v0.1.2

A patch release for the Walrus Memory Python SDK.

Walrus Sites v2.9.2

This patch release fixes how the portal handles SuiNS lookups. When a lookup returns `notExists`, the portal no longer records it as an FN failure. Instead, it stops retrying immediately and resolves the result to `null`. The site-builder also updates its Sui and Walrus dependencies. The release adds no new features, includes no breaking changes, and makes no Move contract changes.

Improvements

• portal: Stop counting `notExists` SuiNS lookups as FN failures; short-circuit retries and unwrap to `null`
• site-builder: Update Sui and Walrus dependencies (Walrus pinned to `f39a84f0` for walrus#3389)

Package v0.1.0

Initial release of the Walrus Memory Python SDK.

Package v0.0.4

Adds `getRememberStatus` for polling async job state and `SealServerConfig` for manual SEAL configuration. Manual mode validates API key pairs.

Package v0.0.1

Initial release of the MCP server with stdio transport, browser wallet login, and inline session tools. Includes memory tools for remember, recall, analyze, and restore.

Walrus v1.48.1

A maintenance release on Mainnet with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Package v0.0.3

Updates `remember()` for the async relayer flow and introduces `rememberAsync`, `waitForRememberJob`, `rememberAndWait`, and bulk remember helpers. Also adds async analysis with `analyzeAndWait`.

Walrus Sites v2.9.1

This patch release fixes portal observability and site-builder configuration. It has no new features, no breaking changes, and no Move contract changes. The site-builder no longer needs the Walrus package ID in `sites-config.yaml` because it now resolves the ID from the Walrus staking object. To pick this up, remove `walrus_package` from your `sites-config.yaml`. Alternatively, update it to the latest package ID from the Walrus `staking_object` in `~/.config/walrus/client_config.yaml`. The portal now counts every request in `ws_num_requests_made_counter`. The site-builder also includes updated Sui and Walrus dependencies.

Improvements

• portal: Count every request in `ws_num_requests_made_counter`
• site-builder: Remove `walrus` `package_id` from `sites-config.yaml` (auto-resolved from the Walrus staking object)
• portal: Wire `bun run typecheck` across `common`, `server`, `worker` and run it in CI
• site-builder: Bump Sui and Walrus dependencies

Package v0.0.2

Initial release of the OpenClaw memory plugin with automatic recall and capture hooks, session summary, CLI commands, and LLM tools.

Package v0.0.2

Adds per-request nonce signing to block replay attacks and replaces delegate key transport with ephemeral SEAL sessions. SDK versions that do not include a nonce are rejected with HTTP 426.

Walrus v1.47.1

Enables storage pool support in the storage node and makes the aggregator return a retryable HTTP 503 `BLOB_UNAVAILABLE` (instead of 500) when a blob is only temporarily unretrievable. Blob attribute headers such as `Content-Type` are now correctly included on range requests when reading by object ID.

Improvements

• enable storage pool support in storage node
• The `GET /v1/blobs/{blob_id}` endpoint now returns
• Blob attribute headers (such as Content-Type) are now correctly included on range request responses when reading blobs by object ID.

Walrus Sites v2.9.0

The site builder now fetches resource dynamic fields with a single batched `multi_get` call. Previously, it used chunked per-resource queries with delays between batches. This change improves performance and reliability when the site builder reads site resources. The `sitemap` subcommand now uses the active wallet environment for its RPC calls. Previously, it used whichever environment appeared first in the config. This release also fixes the script for running the portal locally in Docker and suppresses noisy output from the meta logger.

Improvements

• site-builder: Use multi_get for resource dynamic field fetching
• site-builder: Use active env, not envs.first(), for sitemap RPC
• portal: Fix local Docker portal script and suppress meta logger noise
• dev: Contract upgrade documentation
• site-builder: Bump Sui to testnet-v1.69.1, Walrus to testnet-v1.46.1
• portal: Bump vite from 5.4.21 to 8.0.8 in portal/common

Walrus v1.47.0

The Testnet preview of storage pool support in the storage node, plus the aggregator's new retryable HTTP 503 `BLOB_UNAVAILABLE` response for temporarily unretrievable blobs. Range requests by object ID now return blob attribute headers such as `Content-Type` correctly.

Improvements

• enable storage pool support in storage node
• The `GET /v1/blobs/{blob_id}` endpoint now returns
• Blob attribute headers (such as Content-Type) are now correctly included on range request responses when reading blobs by object ID.

Walrus v1.46.1

Adds a `reuse_resources` query parameter so the publisher and daemon can resume an existing, unfinished upload pipeline instead of starting over.

Improvements

• allow publisher and daemon to add `reuse_resources` query parameters to try to reuse existing (unfinished) upload pipeline

Walrus Sites v2.8.0

This release adds server-side redirects. You define redirect rules with glob pattern matching in `ws-resources.json`. The rules are stored as dynamic fields on the on-chain `Site` object. The portal evaluates redirects before fetching resources and serves a 3xx response. To use redirects, update `package_id` in `sites-config.yaml` to the latest upgraded `walrus_site` package ID. You can also remove the field so MVR resolves it automatically. The portal config field `site_package` is renamed to `original_package_id`. The portal now uses Mysten SDK v2. The server portal's aggregator histogram no longer includes the `blobOrPatchId` and `path` labels.

Improvements

• move/site-builder/portal: Support server-side redirects
• server-portal: Drop blobOrPatchId and path labels from aggregator histogram
• site-builder: Clear checkout credentials before pushing with automerge token
• site-builder: Auto-detect Sui version when bumping Walrus alone
• site-builder: Use automerge token for git push to trigger CI on version bump PRs
• e2e: Build site-builder from main instead of using suiup
• portal: Update portal deps to Mysten SDK v2
• testing: Add a script for testing normal snake operation
• move/site-builder: Update package ids after upgrade

Walrus v1.45.2

Raises the storage node's RocksDB background threads from 2 to 16 and adds the opt-in `garbage_collection.enable_immediate_data_deletion` parameter. Also fixes a combined-daemon body size limit that was silently rejecting uploads with 413 errors and adds a `walrus info coin` CLI subcommand.

Improvements

• Increases the number of RocksDB background threads from 2 (RocksDB default) to 16. Can be configured through the config file.
• Add a new configuration parameter `garbage_collection.enable_immediate_data_deletion`, which defaults to `false`. When processing a `BlobDeleted` event, data is only deleted immediately if this parameter is set to true.
• Fixed body size limit in combined daemon mode (running publisher and aggregator together). The limit was set 1024x too small, causing uploads to be silently rejected with 413 Payload Too Large. Standalone publisher mode was not affected.
• Added a new `info` subcommand to the CLI. `walrus info coin` provides the StructTag of the walrus coin given the CLI's configured context.

Walrus v1.45.1

The Testnet preview that raises RocksDB background threads from 2 to 16 and adds the opt-in `garbage_collection.enable_immediate_data_deletion` parameter. Includes the combined-daemon 413 upload fix and the new `walrus info coin` CLI subcommand.

Improvements

• Increases the number of RocksDB background threads from 2 (RocksDB default) to 16. Can be configured through the config file.
• Add a new configuration parameter `garbage_collection.enable_immediate_data_deletion`, which defaults to `false`. When processing a `BlobDeleted` event, data is only deleted immediately if this parameter is set to true.
• Fixed body size limit in combined daemon mode (running publisher and aggregator together). The limit was set 1024x too small, causing uploads to be silently rejected with 413 Payload Too Large. Standalone publisher mode was not affected.
• Added a new `info` subcommand to the CLI. `walrus info coin` provides the StructTag of the walrus coin given the CLI's configured context.

Walrus v1.44.3

Fixes a storage node crash caused by outdated epoch state, raises RocksDB background threads from 2 to 16, and adds the opt-in `garbage_collection.enable_immediate_data_deletion` parameter.

Improvements

• Fixed an issue that crashed storage nodes due to outdated epoch state.
• Increases the number of RocksDB background threads from 2 (RocksDB default) to 16. Can be configured through the config file.
• Add a new configuration parameter `garbage_collection.enable_immediate_data_deletion`, which defaults to `false`. When processing a `BlobDeleted` event, data is only deleted immediately if this parameter is set to true.

Walrus v1.44.2

Walrus 1.44.2 is a Mainnet release that contains a single storage node fix. It resolves an issue that caused storage nodes to crash because of outdated epoch state. If you operate a Mainnet storage node, this release addresses that crash. The release lists no other changes to storage nodes, clients, or other components.

Improvements

• Fixed an issue that crashed storage nodes due to outdated epoch state.

Walrus Sites v2.7.0

This release improves the reliability of the site builder. Large sites, those with 500 or more routes or resources, no longer fail with `InsufficientGas` errors. These failures came from programmable transaction block (PTB) byte-size overflow, which the site builder now prevents. Site deployment also handles transient Sui object version conflicts better. When a conflict occurs, the site builder automatically retries the operation with exponential backoff instead of failing.

Improvements

• site-builder: Prevent PTB byte-size overflow for large sites
• site-builder: Retry on Sui object version conflicts
• ci: Use walrus-automerge app token and add reviewers to version bump PRs

Walrus v1.44.1

This patch release for Testnet addresses a stability problem in the storage node. Previously, storage nodes could crash when they encountered outdated epoch state. Version 1.44.1 fixes this issue, so nodes no longer crash for this reason. If you operate a storage node on Testnet, this release applies to you. The release notes describe no other changes for developers or for client-side tooling.

Improvements

• Fixed an issue that crashed storage nodes due to outdated epoch state.

Walrus v1.43.1

Allows storage nodes to vote on pricing denominated in USD, a capability that becomes usable once the contract adds support for it.

Improvements

• allow storage nodes to vote on pricing based on USD. This functionality can only be used after the contract supports this.

Walrus Sites v2.6.0

The portal now reads its configuration from a YAML file: `portal-config.yaml` by default, or the path you set in `PORTAL_CONFIG`. All existing environment variables still work as overrides on top of the YAML. See `portal-config.*.example.yaml` for the schema. This release also includes the v2.5.0 priority-based executor. It tries RPC and Aggregator URLs in priority order, with per-URL retries, using the format `URL|RETRIES|METRIC`. `AGGREGATOR_URL` becomes `AGGREGATOR_URL_LIST`, and `RPC_URL_LIST` and `PREMIUM_RPC_URL_LIST` use the new format. The portal still accepts the old format but prints a deprecation warning. The site builder now errors gracefully on invalid glob patterns in `ws-resources.json` and fixes a lowercasing issue with header mappings.

Improvements

• server-portal: Migrate portal configuration from env vars to YAML
• server-portal: PriorityExecutor should fallback based on priority
• server-portal: Add a local-docker-portal.sh script
• site-builder: Error gracefully on invalid glob pattern in ws-resources.json
• site-builder: Fix lowercasing issue with header mappings
• site-builder: Bump script fails when Walrus rev is a tag instead of SHA
• server-portal: Add job for server tests
• site-builder: Add automated Sui + Walrus version bump workflow
• site-builder: Add anyhow error contexts
• server-portal: Add more info to error logs
• portal: Rename priority to metric
• site-builder: Bump walrus & sui deps
• site-builder: Bump bytes from 1.11.0 to 1.11.1
• site-builder: Bump time from 0.3.46 to 0.3.47

Walrus v1.42.1

A maintenance release on Mainnet with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus Sites v2.4.0

This release improves how the Walrus Sites portal reports errors. When an aggregator returns a 403 or 404 error, the portal now handles it and shows clearer messages. If you run a portal, its logs also change. The logger now prints errors correctly, and error log entries include the error's `.cause`.

Improvements

• portal: Handle aggregator 403/404 errors with better messaging
• portal: Print errors correctly in logger
• portal: Add prettier lint workflow for ts files
• portal: Use prettier to format portal
• portal: Include .cause in error log

Walrus v1.41.1

Enables garbage collection by default on Mainnet and adds an alpha `/v1alpha/blobs/{blob_id}/stream` endpoint designed to lower time to first byte when streaming blob data.

Improvements

• Garbage collection is now enabled by default on Mainnet.
• There is a new alpha endpoint `/v1alpha/blobs/{blob_id}/stream` which is intended to reduce latency to first byte, and is intended to be used to stream blob data.

Walrus Sites v2.3.1

This hotfix release disables child process uploads in `site-builder`. That change fixes JSON parsing errors that occur when `site-builder` works with the Walrus CLI. This release requires Walrus CLI v1.37.0 or later. Check your installed Walrus CLI version and upgrade it, if needed, before you use this version of `site-builder`.

Improvements

• site-builder: Disabled child process uploads to fix JSON parsing errors with Walrus CLI
• ci: Simplify dockerhub workflow: use release tag

Walrus v1.41.0

The Testnet preview enabling garbage collection by default and introducing the alpha `/v1alpha/blobs/{blob_id}/stream` endpoint for lower-latency blob streaming.

Improvements

• Garbage collection is now enabled by default on Mainnet.
• There is a new alpha endpoint `/v1alpha/blobs/{blob_id}/stream` which is intended to reduce latency to first byte, and is intended to be used to stream blob data.

Walrus Sites v2.3.0

The site-builder's `deploy --dry-run` output is clearer and now includes Sui gas estimates, so you can preview a deployment's cost before you run it. The site-builder also groups files by size and adds context to the error it reports when site fields are not found. This release fixes updating a site that has expired blobs. On the server portal, it fixes a serialization error that occurred when the portal caught an unexpected exception. It also includes a fix to the portal's common code.

Improvements

• site-builder: Refine 'deploy --dry-run' output, add Sui gas estimates.
• site-builder: Add size-based file grouping.
• site-builder: Add context to the site-fields not found error.
• site-builder: Fix site updating when there are expired blobs.
• server-portal: Fix serializing error when an unexpected exception is caught.
• portal: Fix portal common strict true.

Walrus v1.40.3

Enables database transactions and garbage collection by default on Testnet, both of which can be turned off in the node configuration; operators using `experimental_use_optimistic_transaction_db` should remove that flag.

Improvements

• Enables DB transactions (implemented in [#247](https://github.com/MystenLabs/walrus/pull/247)5) and garbage collection (implemented in [#2542](https://github.com/MystenLabs/walrus/pull/2542), [#2599](https://github.com/MystenLabs/walrus/pull/2599), and [#272](https://github.com/MystenLabs/walrus/pull/272)5) by default on Testnet. The features can be disabled by adding the following to your node configuration:

Walrus Sites v2.2.1

When you update a site, the site-builder now creates new Quilts only for changed resources. Files that stay the same are no longer re-uploaded. The site-builder also uses the correct Walrus package ID from the Staking object when extending blobs, and its error output now includes the config path. The server portal no longer includes the Amplitude integration.

Improvements

• site-builder: Use correct walrus package-id from Staking object when extending blobs
• site-builder: Optimize site updates by creating new Quilts only for changed resources
• site-builder: Add config path to error output
• site-builder: Add Walrus and Seal documentation builder workflow
• site-builder: Add system resource usage tracking to build scripts
• server-portal: Remove Amplitude integration

Walrus v1.39.1

Storage node operators must remove the obsolete `use_legacy_event_provider` flag from their configuration if it is still present.

Walrus Sites v2.1.2

This release fixes a bug in the `site-builder` that affected destroying sites with more than 1000 resources. You can now destroy these large sites with the site-builder. The release also updates the snake example. Its `examples/snake/ws-resources.json` file no longer includes an `object_id` field. If you use the snake example as a starting point for your own site, the updated file reflects this change.

Walrus v1.38.3

Delivers significant memory-usage reductions across the aggregator, publisher, and CLI, plus lower upload latency from fewer Sui RPC requests. Adds a new byte-range read API and an alpha `/v1alpha/blobs/concat` endpoint, and removes the long-unused `get_recovery_symbol` endpoint.

Walrus v1.38.2

The Testnet build with major memory-usage reductions across the aggregator, publisher, and CLI and reduced upload latency. Adds a byte-range read API and the alpha `/v1alpha/blobs/concat` endpoint while removing the deprecated `get_recovery_symbol` endpoint.

Walrus Sites v2.1.1

This bugfix release changes only the Portal component. The server portal now treats only 5xx responses from an aggregator as aggregator failures. Responses outside the 5xx range do not count as aggregator failures. The release does not affect Site Builder. If you use Site Builder to publish or update sites, this release contains no changes for you.

Improvements

• server-portal: Treat only 5xx responses as aggregator failures.

Walrus v1.38.1

An earlier Testnet build of the memory-usage and latency improvements across the aggregator, publisher, and CLI, alongside the new byte-range read API and alpha `/v1alpha/blobs/concat` endpoint.

Walrus Sites v2.1.0

The site-builder can upload files larger than `--max-quilt-size`. It no longer aborts when you update sites with more than 1000 resources. It partly handles invalid object versions returned by a full node, and on Windows it replaces backslashes with forward slashes when converting full paths to resource paths. The portal returns a 503 response when a service is unavailable and a 500 response for unhandled exceptions. The full-node error metric increments correctly, and the portal adds a metric for aggregator fetching.

Improvements

• site-builder: Support files larger than `--max-quilt-size`.
• portal: Enhance error handling with 503 and 500 responses for service unavailability and unhandled exceptions
• site-builder: Fix abort on updating sites with more than 1000 resources.
• portal: Fix full-node error metric not incrementing.
• site-builder: Partly handle invalid object version fetched by full-node.
• site-builder: Replace \ with / in full-path to resource path for Windows.
• portal: Add aggregator fetching metric
• Github Actions: Updates the migrated deploy github action

Walrus v1.37.0

Changes the default consistency check on reads; the previous strict behavior can be restored with the `strict_consistency_check=true` query option or the `--strict-consistency-check` CLI flag.

Walrus Sites v2.0.0

This release removes legacy blob publishing and the `experimental-quilts` feature flag. The portal still serves both Quilt and blob-based sites. Updating a site re-uploads all site data. If that is a concern, test on testnet first, republish to a new site and point SuiNS to it, or stay on mainnet-v1.3.1. The `--check-extend`, `--watch`, `--parallel-stores`, and `--max-concurrent` flags are removed. `update-resource` is renamed to `update-resources`, which accepts multiple `local_path:site_path` pairs. A default 512 MiB limit per Quilt, which also caps individual file size, applies; override it with `--max-quilt-size`. Patch identifiers now use the resource's site path. `StoredQuiltPatch` gains a `range` field for Walrus CLI v1.35 compatibility.

Improvements

• Added comprehensive test coverage for `max-quilt-size` limit enforcement.
• Updated Sui and Walrus dependencies.
• Reduced code duplication in Quilt update tests, improving maintainability.
• Updated github-action, removed --check-extend.

Walrus v1.36.1

Adds a `skip_consistency_check` option (and `--skip-consistency-check` CLI flag) for use only with trusted writers, and trims `server_address` and `server_port` labels from request-duration metrics for fast, successful responses.

Walrus Sites v1.3.1

This release continues work on the `quilts-experimental` feature flag in site-builder. It adds a `DeployQuilts` command that is available only when the `quilts-experimental` flag is enabled. The `update-quilts` command is also now gated under the same flag. Bug fixes in site-builder include respecting the large files limit when using Quilts and inserting routes only after all resources are created. The progress message shown when storing Quilts now reads "and storing Quilts" for clarity. This release also adds tests for the Quilts lifecycle during site updates and for deploying with `--list-directory`.

Walrus v1.35.2

Brings the trusted-writer `skip_consistency_check` option and `--skip-consistency-check` CLI flag to Mainnet, and removes `server_address` and `server_port` labels from metrics for fast, successful responses.

Walrus v1.35.1

Delivers a major reduction in memory usage when reading large blobs and adds the sliver range to `store-quilt` output. Note the warning that the default consistency check will change in v1.37.

Walrus Sites v1.3.0

The site-builder now supports listing directories when you deploy a site with quilts. This release also fixes two bugs in the site-builder. The site-builder no longer deletes a blob when its resource has moved. Quilt deployments now propagate the epoch argument correctly.

Improvements

• site-builder: Support list directory when deploying with quilts
• site-builder: Do not delete moved-resource's blob
• site-builder: Propagate epoch-argument in quilts implementation
• site-builder: Remove temp hack using empty path vec for publish-quilts command
• site-builder: Create new test for duplicate files for quilts publish
• site-builder: Bump versions of site-builder and portal
• vite: Bump vite from 7.0.0 to 7.1.5 in /examples/vite-react-ts
• site-builder: Blob update test with a lot of files

Walrus v1.35.0

The Testnet preview with a major reduction in memory usage for large blobs and sliver-range information in `store-quilt` output, plus advance notice that the default consistency check changes in v1.37.

Walrus v1.34.2

Adds an aggregator `--max-blob-size` option to cap the size of requestable blobs. Note the warning that blobs are stored as deletable by default starting in v1.33; pass `permanent=true` to keep the old behavior.

Walrus Sites v1.2.1

This patch release contains bug fixes for the site-builder and the portal. The site-builder now splits Move commands into programmable transaction blocks dynamically. Dry runs no longer throw an error when you publish with quilts. In the portal, a typo in the logger is fixed, and `getPropertiesWithContext` now passes its arguments correctly.

Improvements

• site-builder: Dynamically split move-commands into ptbs
• site-builder: Fix dry run throwing error for publish quilts
• portal: Fix logger typo, not passing args in getPropertiesWithContext
• Add #[ignore] in integration tests

Walrus v1.34.1

The Testnet build adding the aggregator `--max-blob-size` option, with a reminder that blobs become deletable by default in v1.33 unless `permanent=true` is set.

Walrus v1.33.3

Adds the aggregator `--max-blob-size` option and marks the switch to deletable-by-default blobs; pass `permanent=true` (or `--permanent`) to retain permanent storage.

Walrus v1.33.2

Marks the Testnet switch to deletable-by-default blob storage; add `permanent=true` or the `--permanent` CLI option to keep blobs permanent.

Improvements

• **Warning:** Starting with this release (`v1.33`), blobs are stored as deletable by default; if you want to keep the previous behavior, you need to add the `permanent=true` query parameter.
• **Warning:** Starting with this release (`v1.33`), blobs are stored as deletable by default; if you want to keep the previous behavior, you need to add the `--permanent` option when storing blobs.

Walrus Sites v1.2.0

Header and ignore-file matching in `ws-resources.json` now uses glob patterns, such as `*.md` and `docs/**`, instead of regular expressions. Regex support was never intentional, so if your configuration relies on advanced regex, update it to glob style. This release also fixes how `--list-directory` handles ignore files. Quilt support under the `quilts-experimental` feature expands. The site-builder adds an `UpdateQuilts` command that deletes and re-creates all resources, and it shows quilt patch IDs in publish, update, and sitemap output. The deploy action now supports wallet exported keys and fixes how it uses inputs. Range header building in common-lib is also fixed.

Improvements

• deploy-action: Support wallet exported keys.
• site-builder: UpdateQuilts command, deletes and re-creates all resources.
• site-builder: Use quilt patch ids in site-builder publish/update output.
• site-builder: Show quilt patch ids during sitemap.
• site-builder: Improve the print_summary docs.
• deploy-action: Fix inputs usage.
• site-builder: Use glob instead of regex for ignore files and headers assignments.
• site-builder: --list-directory and ignore files.
• common-lib: nullish coalescing for range header building.
• portal: Integrate Javascript/Typescript tests with Sonarqube.
• deploy: Remove publish_landing_page.sh script.
• site-builder: Update Walrus and Sui dependencies.
• site-builder: Remove unused .with_type() in test-setup resource-fetching.
• site-builder: Create integration test for list directory.

Walrus v1.33.1

An earlier Testnet build introducing deletable-by-default blob storage, with `permanent=true` or `--permanent` available to preserve the previous behavior.

Walrus v1.32.0

Adds an aggregator endpoint, `/quilts/<quilt-id>/patches`, that lists all patches contained in a quilt.

Improvements

• Added new endpoint in aggregator: /quilts/\&lt;quilt-id\&gt;/patches to list all patches in a quilt.
• *Note: This change is an internal code quality improvement and does not affect user-facing functionality. No release notes are required as the changes do not impact the external API or user experience.*

Walrus Sites v1.1.0

This release adds an experimental way to publish sites as quilts instead of blobs. `site-builder publish-quilts --epochs=<EPOCHS> <YOUR-BUILD-DIR>` bundles multiple resources as quilt patches on Walrus, which significantly reduces publishing cost. The command is available only when you build `site-builder` from source with `cargo build --features quilts-experimental`. The portal now derives `quiltpatchid` and retrieves quilt-based resources. `site-builder` now updates the site name when it changes in `ws-resources.json`. An explicit `walrus_package` argument now overrides the config value, and `site-builder` catches transaction failures earlier. The portal and server portal no longer include Sentry or Vercel web analytics.

Improvements

• site-builder: Publish using quilts.
• portal: Derive `quiltpatchid` for resource retrieval.
• portal: Retrieve quilt-based resources.
• portal: Update domain parsing logic and tests for `parse-domain@8.2.2`.
• site-builder: Add more checks for tx failure.
• site-builder: Update the site name if changed in `ws-resources.json`.
• site-builder: Override `walrus_package` from config if passed explicitly as an argument.
• site-builder: Catch errors earlier in site PTB.
• Add workflow to build and publish the server-portal Docker image.
• Create tarball for the site-builder binary in the release workflow.
• Compress the site-builder tarball with gzip and use the `.tgz` extension.
• Add a tarball creation step.
• Add ubuntu-arm64 to the site-builder build list.
• Optionally use `--check-extend` on the site deploy action.
• Clean up portal code by removing the Sentry integration.
• site-builder: Remove unused `network.rs`.
• server-portal: Remove Vercel web analytics and Sentry remnants.
• server-portal: Use mainnet as the default build-arg for suins in Dockerfile.
• site-builder: Remove unused `ResourceOpSummary`.
• portal: Use `toString` in HTML templates to fix build failures.

Walrus v1.31.1

An advance warning that blobs become deletable by default in v1.33; operators who want to keep permanent storage should begin adding the `permanent=true` query parameter now.

Walrus v1.30.1

Storage nodes now restart on a TLS certificate update only when the certificate is near expiry or its subject or extensions have changed, and they emit warning and error logs for nearly expired certificates.

Improvements

• When the TLS certificate is updated, only restart the node if the currently loaded certificate is about to expire or the subject or extensions have changed. Also, add warning and error logs for (almost) expired certificates.

Walrus v1.30.0

The Testnet preview of smarter TLS certificate handling, where a node restarts only on meaningful certificate changes and logs warnings for nearly expired certificates.

Improvements

• When the TLS certificate is updated, only restart the node if the currently loaded certificate is about to expire or the subject or extensions have changed. Also, add warning and error logs for (almost) expired certificates.

Walrus v1.29.2

Brings Quilt batch storage to Mainnet with new publisher, aggregator, and CLI support for storing and reading quilts, plus an `--upload-relay` option. Also fixes startup issues with deleted event databases and lets new nodes start beyond `MAX_EPOCHS_AHEAD`.

Improvements

• Fixes an issue that can occur when a node starts with deleted `events` and `event_blob_writer` DBs.
• New storage nodes can start in epochs later than `MAX_EPOCHS_AHEAD` (53 on the current Testnet and Mainnet) without any additional action.
• Add a new configuration parameter for the event processor config to control the interval at which sampled logs are emitted, with a default of 1 hour (previously 30s).
• Add a `permanent` query parameter. **Warning:** In the future, blobs will be stored as deletable by default; to keep the current behavior, you need to add the `permanent=true` query parameter.
• Added "/v1/quilts" publisher endpoint to store quilts.
• use subsidies contract to proxy register blob call
• Add a `--permanent` option to the `store` and `store-quilt` commands. **Warning:** In the future, blobs will be stored as deletable by default; to keep the current behavior, you need to add the `--permanent` option when storing blobs.
• Add the `--upload-relay` option to the `store` and `store-quilt` commands to allow uploading to Walrus via an [Upload Relay](https://docs.wal.app/operator-guide/upload-relay.html).

Walrus v1.29.1

Introduces Quilt, a batch storage solution for storing and accessing many small files efficiently, with new `store-quilt`, `read-quilt`, and quilt read endpoints across the CLI, publisher, and aggregator. Also fixes deleted-event-database startup issues and relaxes the `MAX_EPOCHS_AHEAD` constraint for new nodes.

Improvements

• Fixes an issue that can occur when a node starts with deleted `events` and `event_blob_writer` DBs.
• New storage nodes can start in epochs later than `MAX_EPOCHS_AHEAD` (53 on the current Testnet and Mainnet) without any additional action.
• Add a new configuration parameter for the event processor config to control the interval at which sampled logs are emitted with a default of 1 hour (previously 30s).
• Add a `permanent` query parameter. **Warning:** In the future, blobs will be stored as deletable by default; to keep the current behavior, you need to add the `permanent=true` query parameter.
• Added "/v1/quilts" publisher endpoint to store quilts.
• Use subsidies contract to proxy register blob call
• Add a `--permanent` option to the `store` and `store-quilt` commands. **Warning:** In the future, blobs will be stored as deletable by default; to keep the current behavior, you need to add the `--permanent` option when storing blobs.

Walrus v1.28.4

Fixes an issue where a blob could be reported as non-existent immediately after certification, affecting both the aggregator and the CLI.

Walrus Sites v1.0.2

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Walrus v1.28.3

Improvements

• Fixes an issue where a blob was sometimes reported as non-existent immediately after certification.

Walrus v1.28.2

Fixes an issue where a blob was sometimes reported as non-existing immediately after certification.

Walrus v1.28.1

Fixes a storage node startup problem with deleted `events` and `event_blob_writer` databases, along with the post-certification blob existence reporting fix.

Walrus Sites v0.1.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Improvements

• site-builder: Setup integration tests for e2e testing.

Walrus v1.27.2

Adds a `CheckpointManager` and a node admin socket for direct operator interaction with a running node, and fixes a crash affecting recovering nodes after restart as well as a checkpoint-lag miscalculation.

Walrus v1.27.1

The Testnet preview adding the `CheckpointManager` component and node admin socket, plus fixes for recovering-node restart crashes and inflated checkpoint-lag values.

Walrus Sites v0.1.0

A maintenance release with no user-facing changes documented in the release notes; see the full commit log for internal updates.

Improvements

• site-builder: Added an expiration column for site resources when running the sitemap.
• site-builder: Introduced `--end-epoch` and `--earliest-expiry-time` flags for defining resource end-epochs.
• site-builder: Implemented a `deploy` command to consolidate the `publish` and `update` commands.
• portal: Added a custom 404 fallback template and integrated it into the URL fetching logic.
• site-builder: Old deletable blobs are now removed after updates.
• site-builder: Updated browsing options based on network context.
• site-builder: Ensured metadata updates when modifying an existing site.
• site-builder: Enforced the `site_name` field and improved the default naming.
• site-builder: Fixed issues with commented values in the configuration.
• portal: Restructured portal logs for better organization.
• contract: Added a README for MVR.
• Added Sonar Analyzer to the CI pipeline for improved code quality checks.

Walrus v1.26.4

Adds multi-URL Sui RPC configuration via `rpc_urls` and removes the implicit fallback to the public Testnet full node. Note the breaking change where metrics and REST API endpoints now bind to the configured address rather than always to `0.0.0.0`.

Walrus v1.26.3

Rate-limits concurrent blob recovery during node recovery and adds multi-URL Sui RPC configuration. Includes the breaking change where services bind to the configured address instead of always to `0.0.0.0`.

Improvements

• feat: rate limit concurrent blob recovery during node recovery
• Use the metrics and REST API address from the configuration file instead of always binding to `0.0.0.0`. Note that this bugfix may break some setups that relied on the previous incorrect behavior. To get the old behavior, set the `metrics_address` and `rest_api_address` to `0.0.0.0:<port>` explicitly.
• Increase default value of `http2_max_pending_accept_reset_streams` parameter to disable warnings from the `h2` crate.
• Reduce verbosity levels of problems with metrics and config sync from ERROR to WARN.
• Allow specifying Sui RPC URL in the configuration file in addition to the CLI option.
• Use the metrics address specified through the CLI options file instead of always binding to `0.0.0.0`. Note that this bugfix may break some setups that relied on the previous incorrect behavior. To get the old behavior, use `--metrics-address 0.0.0.0:<port>` when starting the aggregator.
• Use the metrics address specified through the CLI options file instead of always binding to `0.0.0.0`. Note that this bugfix may break some setups that relied on the previous incorrect behavior. To get the old behavior, use `--metrics-address 0.0.0.0:<port>` when starting the aggregator.
• Add a new `rpc_urls` field to the client config, which allows setting multiple Sui RPC URLs. The Sui RPC URL(s) used for reads are now selected in the following order: (1) CLI argument, (2) client configuration file, (3) wallet config. If none of these are set, the client will **no longer fall back to the `fullnode.testnet.sui.io`** but instead exit with an error.

Walrus v1.26.2

The Testnet build introducing multi-URL Sui RPC configuration and the breaking change where metrics and REST API endpoints bind to the configured address rather than always to `0.0.0.0`.

Walrus v1.25.1

A bug fix adding options to tune compaction write throttling and setting a high write-stalling threshold on the metadata table.

Improvements

• add options to update compaction write throttling, and set high write stalling threshold in metadata table

Walrus v1.25.0

Speeds up batch blob operations by optimizing storage selection, with no change to the user workflow.

Improvements

• Improved performance for batch blob operations by optimizing storage selection. Operations with multiple blobs are now significantly faster with no change to the user workflow.

Walrus v1.24.2

Adds a publisher `force` store option and a CLI `completion` command, and substantially reduces SUI gas costs when registering multiple blobs at once.

Improvements

• Improved documentation for shared blob retrieval.
• Adds the possibility to `force` a store operation, like with the CLI.
• Major reduction of SUI gas cost when registering multiple blobs simultaneously.
• Add new `completion` command to generate completion scripts for various shells.

Walrus v1.24.1

The Testnet preview adding a publisher `force` store option and CLI `completion` command, with a major SUI gas reduction for registering multiple blobs simultaneously.

Improvements

• Improved documentation for shared blob retrieval.
• Adds the possibility to `force` a store operation, like with the CLI.
• Major reduction of SUI gas cost when registering multiple blobs simultaneously.
• Add new `completion` command to generate completion scripts for various shells.

Walrus v1.22.1

Reduces SUI gas costs for simultaneous blob registration, surfaces storage node checkpoint progress in the `health` command, and fixes default-location Sui wallet creation. Operators must remove the `experimental_batch_symbol_recovery` config field.

Improvements

• Remove the `experimental_batch_symbol_recovery` config variable; operators must ensure that the field is not present in their configs.
• Fetch the checkpoint progress info for storage nodes, in the health command.
• Fix a bug that prevented the creation of a Sui wallet at the default location through `walrus generate-sui-wallet`.
• Major reduction of SUI gas cost when registering multiple blobs simultaneously.

Walrus v1.21.2

Improvements

• Optimized default parameters for shard transfer and recovery.

Walrus v1.21.1

Improvements

• Optimized default parameters for shard transfer and recovery.

Walrus v1.20.2

Adds the latest checkpoint sequence number and an estimated checkpoint lag to the CLI `health` output.

Improvements

• Added `Latest checkpoint sequence number` and `Estimated checkpoint lag` in health output.

Walrus v1.18.2

Adds TLS certificate monitoring with automatic reboot, new `--network-key-path` and `--certificate-path` setup options, and a configurable checkpoint request timeout. The publisher's JWT token now includes a `size` field.

Improvements

• Storage node monitors TLS certificate and reboots storage node upon changes.
• Added optional `--network-key-path` and `--certificate-path` options to the `setup` command, removed `--use-legacy-event-provider` and `--disable-event-blob-writer`. Adjusted defaults for storage prices and commission rates.
• Adds a configurable timeout to checkpoint requests.
• Add `size` field of JWTToken

Walrus v1.18.1

The Testnet preview of TLS certificate monitoring with automatic reboot and the new key and certificate setup options, plus the publisher JWT token `size` field.

Improvements

• Storage node monitors TLS certificate and reboots storage node upon changes.
• Added optional `--network-key-path` and `--certificate-path` options to the `setup` command, removed `--use-legacy-event-provider` and `--disable-event-blob-writer`. Adjusted defaults for storage prices and commission rate.
• Add `size` field of JWTToken

Walrus v1.17.2

Adds `--network-key-path` and `--certificate-path` options to the `setup` command, removes legacy event-provider flags, and adjusts default storage prices and commission rates.

Walrus v1.17.1

Improvements

• Fixes a bug where an explicitly set wallet path through the `--wallet` CLI option was ignored.

Walrus v1.16.1

Lets the CLI `delete` command accept multiple files, blob IDs, or object IDs (a breaking change to the delete JSON API) and adds multi-context configuration with a `--context` override. The publisher now keeps created `Blob` objects by default, replacing `--keep` with `--burn-after-store`.

Walrus v1.15.4

Fixes an issue where keys were not loaded into memory when on-chain config parameters were updated, and makes the publisher use the correct default encoding type.

Walrus v1.15.3

Adds support for multiple encoding types across the storage node, publisher, and CLI, and lets the aggregator return allow-listed blob attributes as response headers. Storage nodes can now auto-reload configuration changes from disk, and the CLI gains `--sort-by` and `--desc` options for `health` and `info`.

Improvements

• Automatically reload the StorageNodeConfig from disk and update the on-chain configuration when changes are detected. Some changes may result in an automatic restart of the storage nodes.
• Storage node now supports multiple encoding types. Depending on features selected at compile time, it accepts a subset of them.
• Exposed the aggregator endpoint to allow fetching blobs with associated attributes; if the attribute fields are in the allow-list, they will be included in the response headers.
• Added `encoding_type` query parameter. Added additional variants for the encoding type.
• Added the allowed-headers option for daemon commands, allowing operators to specify custom allowed headers to filter key-value pairs from the blob attribute.
• Added `--sort-by` and `--desc` options for `walrus health` and `walrus info` commands, to sort the storage nodes by node-id, node-name, node-status, or node-url.
• Fix `walrus store` behavior when blobs can be extended.
• The CLI supports multiple encoding types for reads and writes. Depending on features selected at compile time, it uses a different default type and accepts a subset of them for writes.

Walrus v1.14.0

Enables batch symbol recovery by default and adds CLI commands to set, get, and remove blob attributes. The `walrus store` command now extends the lifetime of matching owned blobs instead of re-registering them.

Improvements

• The new batch symbol recovery is enabled by default.
• Added the following new commands for blob attribute:
• The walrus store command now checks for matching owned Blob objects and extends their lifetime if necessary instead of registering the blob again.

Walrus v1.13.2

Fixes the blob-certified metric so it updates only for event blobs and makes the publisher retry when a refresh fails.

Improvements

• fix: Update blob certified metric only if it is event blob
• fix(publisher): retry if refresh fails

Walrus v1.13.1

Reworks TLS configuration by removing the `tls.pem_files` block in favor of `tls.certificate_path` and `network_key_pair.path`, and enables automatic rotation of the protocol key pair. Adds a CLI `--ignore-resources` flag.

Walrus v1.12.0

Lets the `walrus health` command accept a list of node IDs or URLs; this is a breaking change for JSON mode, where `nodeId` and `nodeUrl` become `nodeIds` and `nodeUrls`.

Walrus v1.11.0

Adds a CLI `health` command and the ability to extend owned blobs with WAL, and lets storage node operators change several config parameters by editing the local file and restarting. Deprecates the recovery symbol endpoint in favor of a new one.

Walrus v1.10.0

Makes storage node error responses machine-readable, adds error statuses and richer OpenAPI schemas to the aggregator and publisher, and introduces an optional JWT auth layer for the publisher. Also adds a `walrus-backup` entrypoint and several CLI fixes and options, including staking with multiple nodes.

Improvements

• Update error responses returned by storage nodes to be machine-readable and contain more details
• Add error statuses to the responses of failed requests; update endpoint URLs; and add more detailed OpenAPI schemas.
• Add error statuses to the responses of failed requests; update endpoint URLs; and add more detailed OpenAPI schemas.
• Add optional JWT auth layer for the publisher
• This fixes the cost computation for the `store --dry-run` command.
• Allows staking with multiple nodes simultaneously.
• Fix serialization of JSON output for the `walrus store` command.
• Adds the option to specify `max` as the number of epochs.
• changes to the syntax and output of the `info` command
• an entrypoint called walrus-backup will now be included in the build process. It has minimal (tailing Walrus-related Sui events) functionality at this point.