Honeycomb
Honeycomb bets that the agent, not the engineer, should notice the anomaly first
◆Recent moves
- 1d ago
New: Onboard with Honeycomb MCP
Onboarding moves into the MCP server: connect a codebase, get auto-instrumented through a guided prompt, and query without leaving the coding environment. It attacks instrumentation, the longest-standing barrier to observability adoption, using a surface Honeycomb already shipped.
View source ↗ - 2d ago
Anomaly Detection: Now in Beta
⚡ SPARKThe clearest statement yet of where Honeycomb is heading: detection that runs on its own, feeding the agentic investigation surface the company has been assembling in Canvas.
View source ↗ - 7d ago
Activity Log now includes telemetry stats
Event volume, rejections and errors become a queryable dataset, extending Activity Log from an audit record into a way to monitor the pipeline into Honeycomb itself. Data only accumulates from August 4 forward, so there is no history to look back on yet.
View source ↗ - 13d ago
Edit Triggers, SLOs, and Boards in Canvas
Canvas gains write access to Triggers, SLOs and Boards, with human approval required and automatic investigations explicitly excluded. A deliberate step toward letting the agent change configuration rather than only read it.
View source ↗ - 14d ago
Honeycomb Canvas Connectors: Now in Beta
⚡ SPARKConnectors mark the point where Canvas stops being a Honeycomb-only surface — the arc that both Anomaly Detection and MCP onboarding feed into.
View source ↗ - 18d ago
Authorize multiple teams through OAuth.
MCP clients can authorize against several teams at once with per-team read and write permissions, a practical fix for consultancies and platform teams working across environments.
View source ↗