Errors

The request carried no credentials we could read.

Why: There was no Authorization header, or it did not start with `Bearer `. A common cause is sending the key as a query parameter or a cookie — neither is accepted.

Fix: Send `Authorization: Bearer lnk_live_...` as a request header.

That key is not one of ours, or it is not this key.

Why: The key does not exist, the secret is wrong, or the key failed its checksum. All three answer identically on purpose: telling them apart would let someone probe which keys exist.

Fix: Check for a copy-paste error — a truncated key and a wrong key look the same from here. If you cannot find the key, create a new one; we store only a one-way hash and cannot recover a secret.

This OAuth access token is expired, revoked, or unknown.

Why: Access tokens last one hour. A token also dies the moment the user disconnects your app, which can happen at any time and without warning.

Fix: Exchange your refresh token at `/api/oauth/token`. If that is refused too, the user has disconnected you — send them through the consent screen again.

This key passed the expiry date set when it was created.

Why: The key had an expiry and that date has now passed.

Fix: Create a replacement key. Next time, roll the key before it expires — a roll gives you the new secret immediately and keeps the old one working for a grace period, so nothing goes down.

This key was revoked and will never work again.

Why: Somebody revoked it in the dashboard, or it was revoked automatically because it was found in a public code repository.

Fix: Create a new key. If it was revoked for being public, remove it from your repository history as well — not just the latest commit — and put the replacement in an environment variable.

This key is valid, but it was not given this permission.

Why: Permissions are per key and checked per request. The message names the exact one that is missing.

Fix: Edit the key in the dashboard and tick the named permission, or use a key that already has it.

The user did not grant your app this permission.

Why: A user can untick individual permissions on the consent screen, so a scope you requested is not necessarily a scope you were given.

Fix: Read the `scope` field returned with the token rather than assuming you got what you asked for. Send the user back through the consent screen if you genuinely need the extra permission.

This key is limited to certain resources, and this is not one.

Why: The key has a resource scope — a set of domains, a set of tags, or both. Domains and tags are combined, so a key scoped to both may only touch links matching both.

Fix: Use a key without that restriction, or widen the scope in the dashboard. The message names the current scope.

This key may only be used from specific IP addresses.

Why: Either your address is not on the key's allow-list, or we could not read your address at all. The second case is also a refusal: a restriction that stops applying when the address is unreadable is not a restriction.

Fix: Add the calling server's address to the key's allow-list, or use a key with no IP restriction. Note that your outbound address is often not the one you expect on serverless platforms.

This key may only be used during set hours.

Why: The key has a schedule. The window is evaluated in the timezone stored on the key, which is not necessarily yours or the server's.

Fix: Wait for the window, or change the schedule in the dashboard. The message names the window and its timezone.

This key is paused, so every request is refused.

Why: Somebody paused it, or we paused it automatically because it was used from a different continent than usual. The second is a heuristic and it does produce false positives — a new server, a VPN, a trip.

Fix: Resume the key in the dashboard if you recognise the activity. If you do not, roll it instead: that issues a new secret and stops the old one.

This endpoint cannot be used with an API key at all.

Why: Key management, billing and the OAuth endpoints are session-only. A key that could create keys could issue itself full access and outlive its own revocation.

Fix: Do this in the dashboard while signed in.

Your role in this workspace does not allow this action.

Why: Some actions are limited to owners and admins. A key authenticates as the workspace, so this usually means a session with a viewer or member role.

Fix: Ask an owner or admin of the workspace to do it, or to change your role.

You are not a member of the workspace you named.

Why: The `X-Workspace-Id` header named a workspace you do not belong to.

Fix: Drop the header to use your default workspace, or send an id you are actually a member of.

This page is waiting for approval.

Why: The workspace requires a review before anything is published. The message names what is still outstanding.

Fix: Ask a reviewer to approve the page, or turn the approval requirement off in the workspace settings if it is not wanted.

This page cannot be cloned.

Why: Its owner never listed it as clonable, or it has since been withdrawn from the gallery. Cloning is opt-in, not the default.

Fix: Only gallery pages whose owner opted in can be cloned. Build from a template instead.

An administrator locked some of the fields you tried to change.

Why: Locked fields are resolved at render time from a shared template, so a change here would be silently overwritten on the next render anyway. Refusing is more honest than accepting and discarding it.

Fix: The message names the locked fields. Change them on the template itself, or ask an admin to unlock them.

One or more fields in the request body were not acceptable.

Why: A field is missing, the wrong type, or outside its allowed range.

Fix: Read `error.details.fields`. It is keyed by field name and names the problem with each one.

Extra detail: `fields`: an object mapping each bad field name to its problems.

The request body was not valid JSON.

Why: Usually a trailing comma, a single-quoted string, or a body sent with no `Content-Type: application/json` header at all.

Fix: Serialise the body with a real JSON encoder rather than string concatenation.

We will not shorten that destination.

Why: The URL uses a scheme we do not allow, points at a private network address, or matched a malware or phishing list.

Fix: Use a public `http:` or `https:` URL. If you believe the block is wrong, contact support with the exact URL.

That short link already exists on that domain.

Why: Slugs are unique per domain, not globally. The same slug can exist on two different domains.

Fix: Pick another slug, or omit `slug` entirely and we will generate an unused one.

The barcode could not be produced from those parameters.

Why: Each symbology has strict rules about length, character set and check digits. EAN-13 needs exactly thirteen digits; Code 39 refuses lowercase.

Fix: Read the message — it names the rule that failed — and correct the data or pick a symbology that can encode it.

The barcode parameters are not usable.

Why: Either no symbology was named, or the data contains characters the named symbology has no way to encode.

Fix: Name a supported symbology and check the data against its character set and length rules.

This collection needs a title before it can be created.

Why: The title field was missing, or present but empty. A whitespace-only value counts as empty.

Fix: Send a title with at least one non-whitespace character in it.

That page slug is already in use.

Why: Page slugs are unique within a workspace. An archived page still holds its slug, which is the case people do not expect.

Fix: Pick a different slug, or delete the page holding it. Archiving is not enough to free one.

That approval action does not apply to the page's current state.

Why: Approval moves through fixed states in order. You cannot approve a page that was never submitted, or submit one that is already approved.

Fix: The message names the current state. Pick the action that follows it, or read the page's approval history to see how it got there.

The template pack is not complete enough to publish.

Why: A pack needs a name and at least one template before anybody could usefully install it. The message names which of those is missing.

Fix: Fill in what the message names, then publish again. The pack is saved as a draft in the meantime.

An A/B test is already running on this page.

Why: Two overlapping tests split the same traffic, and neither result can then be attributed to either change.

Fix: End the current test first. Its results stay available after it ends, so nothing is lost by stopping it.

The UTM values break this workspace's naming convention.

Why: The workspace enforces a convention so campaign reporting stays consistent across everyone using it. The message names the rule that failed.

Fix: Follow the convention, or relax it in the workspace settings if it is getting in the way more than it helps.

That is not a valid web address.

Why: The value could not be parsed as a URL at all — usually a missing scheme or a stray space.

Fix: Send a complete URL including the https:// prefix, and trim any surrounding whitespace.

We will not fetch that address.

Why: It resolves to a private, loopback or link-local address, or it does not resolve at all. Fetching those would let a request reach networks it has no business in — including our own internal ones.

Fix: Use a publicly reachable address. A tunnel service works if you need to expose something running locally.

The page did not answer successfully.

Why: It returned an error status, which the message includes. A 403 usually means a bot filter; a 404 usually means a typo.

Fix: Open the address in a private browser window without signing in. If it fails there too, that is what we are seeing.

We found no links on that page.

Why: The page builds its links in the browser after load. We read the HTML exactly as the server sends it, and there were no links in it yet.

Fix: Add the links by hand, or import from a page that renders its links server-side.

The endpoint is paused, so a replay would go nowhere.

Why: An endpoint pauses itself after repeated failures. Replaying into a paused endpoint would fail again and count as another failure, pushing it further from recovery.

Fix: Send a test ping to confirm the receiver is healthy, switch the endpoint back on, then replay.

That affiliate is already enrolled on this page.

Why: A second enrolment would split their attribution across two rows, so neither would show their real earnings.

Fix: Nothing to do — they are already enrolled and their existing referral code still works. Update the existing enrolment if you meant to change its terms.

You own a workspace that other people are in.

Why: Deleting the account would take the workspace and everyone else's work with it. We will not do that on one person's say-so.

Fix: Transfer ownership of the workspace named in the message to another member, then request deletion again.

The erasure request cannot proceed yet.

Why: Something has to be resolved first — usually an active subscription, or a workspace you own that other people are in. The message names it.

Fix: Resolve what the message names, then request erasure again. Nothing has been deleted in the meantime.

A feature flag with that key already exists.

Why: Flag keys are unique within a workspace, because evaluation looks a flag up by key and two matches would make the result arbitrary.

Fix: Update the existing flag rather than creating a second one, or pick a key that is not taken.

No flag key was given to evaluate.

Why: The endpoint evaluates named flags and was given no names, so there is nothing for it to answer.

Fix: Add ?key=your-flag-key to the query string. Repeat the parameter to evaluate several flags in one request.

The documentation playground would not run that request.

Why: A path parameter was left empty, or a destructive call was not confirmed. The playground only runs endpoints that appear in this reference, so it can never reach one the documentation does not describe.

Fix: Read the message — it names the specific reason — and fill in what it asks for.

The Ryabi-Version header names a version we do not have.

Why: Usually a typo. An unknown version is refused rather than ignored, because a pin you believed in that silently did nothing is worse than no pin at all.

Fix: Use one of the versions listed on the changelog page, or leave the header off entirely to track the current version.

No page content was sent to link.

Why: The body had no `content` field, or it was empty. There is nothing to search for keywords in.

Fix: Send the page as `{ "content": "<p>…</p>" }`.

`format` must be auto, html or markdown.

Why: An unrecognised value. It is refused rather than guessed, because writing Markdown link syntax into HTML — or the reverse — corrupts the file rather than producing a wrong link.

Fix: Leave `format` off to detect it, or send exactly html or markdown.

`at` is not a date we can read.

Why: `at` pins the moment scheduled rules are judged against, so a build that runs either side of a campaign boundary produces the same output. An unreadable value is refused rather than silently treated as now.

Fix: Send an ISO 8601 date-time, for example 2026-12-31T23:59:59Z.

Too many requests in the last minute.

Why: Limits are per key, per minute. A key with its own allocation gets exactly that allocation and no more, even when the rest of the workspace is idle.

Fix: Back off and retry — exponentially, not in a tight loop. If you need more throughput for one integration, give that key a larger allocation in the dashboard.

This key used its whole monthly call budget.

Why: The key has a monthly cap and has reached it. It resets at the start of the next calendar month, in UTC.

Fix: Raise the cap in the dashboard, or wait for the month to roll over. If this was unexpected, check for a retry loop — that is what the cap is there to catch.

Your plan's total link allowance is used up.

Why: This counts every link in the workspace, not links created through the API. Links created by a test-mode key do not count.

Fix: Delete links you no longer need, or upgrade the plan.

Your plan's allowance of saved QR codes is used up.

Why: This counts SAVED standalone QR codes only. A QR attached to a short link does not count — that link is already metered by the link allowance — and a one-off render from /v1/qr/render is free and counts against nothing. QR codes saved by a test-mode key do not count.

Fix: Delete a saved QR code you no longer need, or upgrade the plan.

You asked for QR codes on a page of more than 50 links.

Why: `include=qr` renders a real SVG per link. Fifty is already a lot of work for one request; a few hundred is a request that dies half-way and tells you nothing, so it is refused up front instead.

Fix: Lower perPage to 50 or below, or drop include=qr and fetch codes for the links you actually need.

Too many CLI listen sessions are open at once.

Why: There is a per-workspace cap, named in the message. Abandoned terminals hold their slot until the session times out.

Fix: Close a session you are no longer using, or wait for an idle one to expire on its own.

The page is bigger than this endpoint will process.

Why: Keyword matching holds the whole document in memory and scans it once per rule, so there is a ceiling on how large one page may be.

Fix: Split the page, or link the pieces separately as your build renders them. A generous article is well under the limit.

No such resource in this workspace.

Why: The id does not exist, or it belongs to a different workspace. Both answer identically so an id cannot be probed from outside.

Fix: Check the id, and check you are acting in the right workspace. If you are using a test-mode key, remember it cannot see production resources.

No such link in this workspace.

Why: As above. The most common surprise is a test-mode key looking for a live link, or the reverse — the two are separate sets.

Fix: Check the id and the key's mode.

The signed-in account has no matching record.

Why: Rare, and usually means a half-finished signup.

Fix: Sign out and back in. If it persists, contact support.

This account does not belong to any workspace yet.

Why: Signup completed but no workspace was created.

Fix: Create a workspace in the dashboard, then retry.

No such QR code in this workspace.

Why: The id does not exist, or it belongs to a different workspace. Both answer the same way so an id cannot be probed from outside.

Fix: List your QR codes first and use an id from that response. If you are sending an X-Workspace-Id header, check it names the right workspace.

No such menu in this workspace.

Why: The id does not exist, or it belongs to a different workspace. Both answer the same way so an id cannot be probed from outside.

Fix: List your menus first and use an id from that response. If you are sending an X-Workspace-Id header, check it names the right workspace.

None of those link ids are in this workspace.

Why: Every id was wrong, or they all belong to a different workspace. A test key looking for live links produces this too — the two sets are separate.

Fix: List the links first and take the ids from that response. Check the key's mode if you are working with test data.

Some of those link ids are not in this workspace.

Why: A partial match. The message says how many of the ids were found, which separates a single typo from a whole batch aimed at the wrong workspace.

Fix: Nothing was changed — the operation is refused whole rather than applied to the ids that did match. Correct the missing ids and send it again.

No such page in this workspace.

Why: The id does not exist, or it belongs to a different workspace. Both answer the same way so an id cannot be probed from outside.

Fix: List your pages first and use an id from that response. If you are sending an X-Workspace-Id header, check it names the right workspace.

No such template or template pack.

Why: The id is wrong, or the template has since been deleted or unpublished by its owner.

Fix: List the available templates first and use an id from that response rather than one you stored earlier.

The page is not published.

Why: Analytics and other public surfaces only exist once a page is live, because nothing can have visited a page that was never served.

Fix: Publish the page first. Analytics start from the moment it goes live, not from when it was created.

That CLI listen session is not open.

Why: It timed out, or it was closed from another terminal. Sessions expire on their own so an abandoned one cannot hold a slot forever.

Fix: Start the session again. Any events sent while it was closed are lost.

No account uses that email address.

Why: An affiliate needs an account before they can be enrolled, because there is nothing to attribute earnings to or pay out to otherwise.

Fix: Ask them to sign up first, then enrol them with the same address they signed up with.

API access is not included on this workspace's plan.

Why: The API is a paid feature. The check is against the workspace's plan, not your personal one.

Fix: Upgrade the workspace to Pro or higher.

That specific feature is not on this plan.

Why: The request was valid, but it used a field gated behind a higher plan — A/B testing and some routing options, for example.

Fix: Remove the gated field, or upgrade. The message names the feature.

The rate limits handed out to your keys add up to more than your plan allows.

Why: Allocations are slices of one workspace-wide ceiling. The total across all keys cannot exceed it.

Fix: Lower another key's allocation first. `error.details` gives the current total and the ceiling.

Extra detail: `allocated`: current total. `ceiling`: what the plan allows.

This feature needs a paid plan.

Why: The feature is gated behind a plan the workspace is not on. The message names which plan it needs.

Fix: Upgrade the workspace, or remove the gated option from the request if you do not need it.

Webhooks are not included on this workspace's plan.

Why: Webhooks are a Business feature. The check is against the workspace's plan, not your personal one.

Fix: Upgrade the workspace to Business or higher. Polling the API is a workable stand-in until then.

That domain is already connected somewhere.

Why: Either it is already on your workspace, or another workspace claimed it first. The message distinguishes the two, because a domain can only be connected once across the whole platform.

Fix: If it is already yours, there is nothing to do. If another workspace has it and the domain is genuinely yours, contact support with proof of ownership.

That is not a domain name.

Why: The value has a scheme, a path, a port, or characters a hostname cannot contain. Pasting a full URL instead of a hostname is the usual cause.

Fix: Send just the hostname — example.com, not https://example.com/path. Trim any trailing slash as well.

That domain cannot be used to publish a page.

Why: Some hostnames are reserved for the platform itself. Publishing on one would let a page impersonate part of our own site, which is a phishing vector regardless of who asked for it.

Fix: Publish on your own connected domain, or on the shared link domain.

The domain is connected but not verified yet.

Why: Verification proves you control the DNS. Until it passes we will not serve anything on the domain, because serving on an unverified name is how a domain takeover becomes a hosted phishing page.

Fix: Add the DNS records shown in the dashboard, then wait for them to propagate. That is usually minutes and occasionally an hour.

The forwarding target is not a valid URL.

Why: A domain forward needs a complete, absolute URL. A bare hostname or a path on its own gives a browser nowhere to go.

Fix: Send a full URL including the scheme, for example https://example.com.

That forward points the domain at itself.

Why: A domain forwarding to its own hostname never terminates. The browser gives up after a handful of hops and shows the visitor an error page.

Fix: Point the forward at a different hostname. A loop is invisible to a target probe, which is why it is refused here rather than discovered later.

No file was attached.

Why: The request had no field named file. Sending a JSON body to an upload endpoint produces this too, because there is no file field to find in one.

Fix: Send the request as multipart/form-data with a field named file. Do not set Content-Type by hand — your HTTP client has to add the boundary.

No files were attached.

Why: This endpoint takes a repeated files field and received none of them.

Fix: Attach at least one entry named files. Repeat the field name once per file rather than sending an array in a single field.

More files than this endpoint accepts.

Why: There is a per-request cap so one upload cannot exhaust the request timeout. The message names the exact limit.

Fix: Split the upload across several requests. Nothing was stored, so there is nothing to clean up before retrying.

The file was zero bytes.

Why: Usually a form field that was set but never populated, or a read stream that was already consumed before the request was built.

Fix: Check the file's size on disk before uploading. If you are streaming, make sure the stream has not already been read.

The file is over the size limit.

Why: There is a fixed maximum per file, named in the message. It applies to the encoded upload, which is slightly larger than the file on disk.

Fix: Compress the file, or host it somewhere else and link to it instead of uploading it here.

The media file is too large to transcribe.

Why: Transcription has its own ceiling, lower than ordinary uploads, because the model provider imposes one.

Fix: Trim the recording, re-encode it at a lower bitrate, or paste a transcript directly if you already have one.

We could not download the media you pointed at.

Why: The URL timed out, refused the connection, or answered with something that was not a media file. A link requiring a login does this.

Fix: Check the URL is publicly reachable, then try again.

That file type is not allowed.

Why: Executables, scripts and archives that can contain them are refused. A storefront that can serve an .exe is a malware distribution channel with our name on it.

Fix: Package the file as a PDF or a zip of documents, or host it yourself.

That file type is not one this endpoint handles.

Why: Different endpoints accept different types, and the message names the one it actually received — which is detected from the bytes, not from the filename.

Fix: Convert the file to a supported format, or use the endpoint that handles this type.

The image could not be decoded.

Why: The bytes are not a format we can read, or the file is truncated. Renaming a file to .png does not make it one, and that is the usual cause.

Fix: Open the file to confirm it is really an image, then re-export it as PNG, JPEG or WebP and upload again.

The image edit could not be applied.

Why: Either the source file is unreadable, or the crop rectangle falls partly or wholly outside the image.

Fix: Read the image dimensions first and clamp the crop to them. Nothing was stored, so the original is untouched.

We could not store the file.

Why: A storage failure on our side. Nothing about your request caused it, and nothing was partially written.

Fix: Retry once after a short delay. If it keeps happening, contact support with the time — that is enough for us to find it.

This endpoint expects multipart/form-data.

Why: A JSON body was sent to an endpoint that takes a file upload, so there was no form data to read.

Fix: Send the request as multipart/form-data. Let your HTTP client set the Content-Type so it can add the multipart boundary.

The expiry date is not usable.

Why: It could not be parsed, or it is already in the past — which would make the thing you are creating unavailable the moment it existed.

Fix: Send an ISO 8601 date-time in the future, for example 2026-12-31T23:59:59Z.

The download limit is not a positive number.

Why: Zero and negative values would make the file unavailable the moment it was created, which is never what anyone means.

Fix: Send a whole number of one or more. Omit the field entirely if you want unlimited downloads.

That product has sold out.

Why: Stock is checked when checkout opens, not when the page rendered, so a page loaded a few minutes ago can still hit this. That is deliberate — the alternative is overselling.

Fix: Restock the product in the dashboard, or remove the stock limit if it was not intended. No payment was taken.

The coupon cannot be used.

Why: It does not exist, it has expired, it has been redeemed its maximum number of times, or it does not apply to the items in this order.

Fix: Check the code's conditions in the dashboard. All four causes answer the same way so a coupon code cannot be brute-forced from the response.

The cart contains items priced in different currencies.

Why: One checkout is one currency. We will not guess an exchange rate on somebody's behalf and charge them the result.

Fix: Split the order, or price the items in a single currency.

That slot has no price yet.

Why: A bookable slot cannot be sold until a price is set on it. A slot with no price would otherwise check out at zero.

Fix: Set a price on the slot in the dashboard, or mark it as free if that is what you meant.

This booking is not priced correctly.

Why: Its pricing configuration is incomplete or inconsistent, so no checkout can be built from it without guessing an amount.

Fix: Open the booking in the dashboard and finish its pricing. The visitor was not charged.

Checkout is not available right now.

Why: The payment provider is unreachable, or it is not configured for this workspace. No payment was attempted either way.

Fix: Try again shortly. If it persists, check the workspace's payment settings are complete.

The payment provider refused to open a checkout.

Why: The provider returned an error rather than a checkout session. The message passes on what it said, verbatim.

Fix: Read the provider's message — it is usually specific. If it is not actionable, contact support. No payment was taken.

This plan has no price configured yet.

Why: A plan needs a price registered with the payment provider before it can be bought. This is a gap in the deployment's configuration, not a fault in your request.

Fix: Contact support — no change to your request will help, because the missing piece is on our side.

This workspace already has an active subscription.

Why: Opening a second checkout would create a second subscription and bill the workspace twice for the same thing.

Fix: Use the billing portal to change plans instead. Upgrades and downgrades are prorated there.

This AI feature is not configured on this deployment.

Why: AI features need a model provider key. Without one they ship dark rather than half-working, so the endpoint says so plainly instead of failing in a way that looks like a bug.

Fix: An operator has to configure the provider. Nothing you change in the request will help.

The model could not complete the request.

Why: The provider returned an error or timed out. Usually transient, and usually unrelated to what you sent.

Fix: Retry once after a short delay. If it keeps failing on the same input, that input is likely the problem — try a smaller or simpler one.

The source image or URL could not be used.

Why: The message names the specific problem: unreachable, the wrong type, or over the size limit.

Fix: Read the message and correct the source. If the URL needs a login, upload the file directly instead.

There was nothing readable on that page.

Why: The page builds its content in the browser, or it is behind a login. We read the HTML exactly as the server sends it, and that HTML was empty of anything usable.

Fix: Write the title and description by hand, or point at a page that renders its content server-side.

The generated QR artwork could not be scanned.

Why: Every candidate is decoded before it is offered to you, and none of them decoded. A busy or dark style destroys the contrast a scanner needs. Offering a pretty code that does not scan would be worse than refusing.

Fix: Try a simpler prompt, a lighter style, or fewer visual elements around the code's corners.

That endpoint does not accept this HTTP method.

Why: Usually a GET where a POST was meant, or the reverse.

Fix: Check the reference for the method this endpoint expects.

The QR code could not be rendered.

Why: A rendering failure on our side, most often from an unusual combination of size, error-correction level and style.

Fix: Retry with the default options. If that succeeds, reintroduce your options one at a time to find the one that breaks it.

The share card could not be rendered.

Why: An image-generation failure on our side. Nothing about your request caused it, and nothing was stored.

Fix: Retry once after a short delay. If it persists, contact support with the page id and the time.

The code could not be verified.

Why: Something failed while decoding. This is not a decision that the code is invalid — that would come back as a successful response saying so.

Fix: Retry once. A failure that persists on the same image is worth reporting with the image attached.

The affiliate could not be enrolled.

Why: A failure on our side rather than a problem with your request. The enrolment was not partially created.

Fix: Retry once after a short delay. If it persists, contact support with the page id and the affiliate's email.

A referral code could not be assigned.

Why: A failure on our side while generating a code that is not already taken. The affiliate was not left half-enrolled.

Fix: Retry once after a short delay. If it persists, contact support with the page id.

This feature is not enabled on this deployment.

Why: It needs a third-party credential that has not been configured. Features like this ship dark rather than half-working, so the endpoint says so rather than failing in a way that looks like a bug.

Fix: An operator has to configure the credential. Nothing you change in the request will help.

The music resolver could not find or reach that release.

Why: Either the release is unknown to the provider, or the provider itself is unavailable. The message distinguishes the two, because only one of them is worth retrying.

Fix: Add the streaming links by hand if the release is unknown. Retry if the provider was down.