errorcore docs
Stack traces tell you where. errorcore tells you why.
When code breaks in production, the errorcore SDK captures the state that led to the failure: local variables in each frame, the ordered I/O that ran before the throw, the request that triggered it, and the async trace context. It scrubs sensitive fields, encrypts the package inside your process, and ships it to the errorcore ingestion API. The backend authenticates the envelope, stores it as received, decrypts it in a worker, and groups identical failures into a case you read in the console.
These docs cover the parts you operate: installing and configuring the SDK, the credentials it needs, what the ingestion API accepts, how usage is counted, and how agents read your incidents.
Start here
- Get your first errorcore event: the five-minute path.
- Install the SDK
- Use a coding agent: one prompt does the whole setup.
- Environment variables
Credentials
errorcore uses two distinct server-side secrets, and neither one is a license key.
- API keys:
ec_live_bearer credentials that authenticate your servers to the ingest API. - Encryption keys: the
ERRORCORE_DEKyour process uses to encrypt payloads.
Operate
- Billing and usage: plans, hard caps, and quota versus rate limiting.
- Ingestion pipeline: what happens after an envelope arrives.
- Agent access: read-only MCP access for coding agents.
- Security and PII
- Troubleshooting
- Production checklist