Planned
The honest list of what Sypnia does not do at launch, in three groups: what is planned, what is absent with no promise, and what Sypnia will not do by design. No dates.
This page is for anyone who expected something and did not find it, and for coding agents, so that they never describe a planned thing as if it existed. It covers the open-source engine Engine and Sypnia Cloud Cloud.
Everything else in these docs describes what exists at launch. This page is the one place for what does not. It has no dates, on purpose: a date we miss would be a promise we broke.
| Group | What it means |
|---|---|
| Planned | We intend to build it. In the docs it carries the Planned badge. It is not started, or not finished. |
| Not in the first version | It does not exist at launch, and we make no promise either way. |
| Not planned, by design | A decision. Do not wait for it. |
When something ships, it leaves this page and appears in the changelog.
Planned
SDKs and platforms
| Planned | What to use until then | Described in |
|---|---|---|
| Native iOS and Android SDKs | Mobile apps built as web views (Capacitor, Cordova, a wrapped web app) work at launch with the browser SDK. A fully native app has no SDK. | Browser SDK |
| SDKs for Python, PHP, Go and Rust | Send traces from any backend with OpenTelemetry, over OTLP/HTTP. | OpenTelemetry |
An adapter for AWS Lambda handlers that receive API Gateway or function URL events, in @sypnia/node | The OpenTelemetry intake. Fetch-style functions are covered at launch by wrapFetchHandler. | Node SDK |
Bun, Deno and Cloudflare Workers as tested runtimes for @sypnia/node | wrapFetchHandler uses only web-standard APIs, so it may work. It is not tested there. | Node SDK |
More database drivers traced by themselves in @sypnia/node: MySQL, MongoDB, Redis, Prisma’s own engine | Wrap the call in span(), or use the driver’s OpenTelemetry instrumentation. pg and Supabase calls are traced at launch. | Node SDK |
The Edge runtime in Next.js: routes with runtime = 'edge', and middleware | Nothing. That code is not traced and its errors are not reported. | Next.js SDK |
| Source maps for Next.js server bundles | A server error’s stack shows the built file. Its route, its trace and the page that showed it are there. | Next.js SDK |
| Sampling of successful server requests | Every request is traced. Leave routes out with ignoreRoutes. | Sampling and quotas |
Kinds of data
| Planned | What to use until then | Described in |
|---|---|---|
| Logs | Sypnia does not collect logs. OTLP logs are refused with 404. | OTLP |
| Custom metrics and dashboards | Custom events with track(name, props) from the browser, counted per day. For your own charts, read the API or the daily Parquet files of a self-hosted engine. | What Sypnia is |
Access and identity
| Planned | What to use until then | Described in |
|---|---|---|
| Single sign-on (SSO) Cloud | Sign in with e-mail and password, or with GitHub. | Organizations and teams |
| A signed identity for sending, for apps that need proof of who is sending events | The public ingest key is the only credential for sending. Allowed origins, rate limits and quotas protect it. | Projects, keys and tokens |
Not in the first version
These do not exist at launch. They are listed so that you do not look for them, not because they are promised.
| Not at launch | What there is instead |
|---|---|
Custom events from the server. track() exists in the browser SDK only. | Send the event from the browser, or wrap the step in span() with attributes. |
| A warning before a certificate expires, in uptime checks. | An expired certificate fails the check, and the «App down» alert says why. |
| A public status page for your app, fed by uptime checks. | GET /v1/uptime/checks has the data to build one. |
| A public status page for Sypnia Cloud, a guaranteed response time, a service level agreement. Cloud | Owners of affected organizations are told by e-mail. See Support. |
| Signing in to the MCP server through the browser (OAuth). | An API token in the client’s configuration. See MCP authentication. |
| Identifying people from OpenTelemetry attributes. | A trace takes the person of its session. See Identifying people. |
| Tags, contexts and extra data on an error, and searches or saved views over them. | An error carries its session, person, account, release, route and trace. See Coming from Sentry. |
One session across several origins, such as shop.example and app.shop.example. | Each origin has its own sessions. |
| Sub-teams, or access per project inside an organization. Cloud | Every member sees every project. Use two organizations. |
| Two-step sign-in of Sypnia Cloud’s own. Cloud | Sign in with GitHub and use GitHub’s. |
| A choice of region, and encryption keys that you manage. Cloud | Everything is in us-east-1. To keep data elsewhere, self-host. |
| A one-click export of a project. Cloud | Everything can be read through the API while it is kept. |
| A public API for users, organizations, plans and billing. Cloud | The app. |
| Pull requests on GitHub Enterprise Server, GitLab or Bitbucket. Cloud | The prompt, or the MCP server, with your own coding agent. |
| Deleting a single problem, or only a replay, through the API. | Ignore the problem; delete the session to remove its replay. See Deleting data. |
| Retention per project on a self-hosted engine. Engine | One set of retention settings for the whole engine. |
| Uptime checks from several places on a self-hosted engine. Engine | Checks run from where your worker runs. Sypnia Cloud checks from three regions. |
| More screens in the Console. Engine | The Console shows problems and replays. The rest is in the API, the CLI and the MCP server. See The Console. |
| A measured cost and capacity table for the engine. Engine | Estimates with their arithmetic. See Scaling and cost. |
Not planned, by design
These are decisions, not gaps.
| Sypnia will not | Because | What there is instead |
|---|---|---|
| Send alerts to WhatsApp or Slack, in the engine or in Sypnia Cloud | Alerts are e-mail and in the app on Sypnia Cloud, and a webhook everywhere. | The alert webhook, and your own function on it. See Alerts and webhooks. |
| Put an e-mail sender in the engine’s default deployment | The default deployment holds only what cannot be left out. | A recipe: Alert e-mails from a self-hosted engine. |
| Open pull requests from the engine | Pull requests need an app installed in your repository, which is part of Sypnia Cloud. | The ready prompt and the MCP server, with your own coding agent. |
| Merge code, or push to your default branch | A person reviews and merges. | Pull requests. |
| Add users and organizations to the engine | The engine knows projects, keys and tokens. Accounts are Sypnia Cloud. | One API token per person. |
| Record what a person types, or let it be unmasked | It is the one thing that must never leave the browser. | Masking. |
| Let the agent read replay recordings | The analysis works from events, stacks and traces. | Models and what is sent. |
| Promise a specific AI model | Models change. On Sypnia Cloud we choose and change them; on the engine you choose. | Models and what is sent. |
| Require a service besides Postgres and a bucket to run the engine | Two stores are the design. | How the engine is built. |
Ask for something
- For the open-source engine and the SDKs: open an issue in the repository and say what you need it for. See Contributing. Several things on the planned list are good places to help.
- For Sypnia Cloud: write to
support@sypnia.com.
Asking does not move a date, because there are none. It does tell us what to build first.
For coding agents
If you are an AI agent reading these docs: nothing on this page exists at launch. Do not write code that calls it, do not tell a person to configure it, and do not present it as available. Offer the alternative in the middle column instead.