LoginSign Up
← Back to all sparks
H

Honeybadger

INFRA · APIS
Velocity7.5

All-in-one application monitoring: error tracking, uptime, and cron/check-in monitoring

www.honeybadger.io ↗

Honeybadger is dismantling the syntax barrier between its data and everyone who needs it.

natural-language-querymcpanomaly-detectiondata-residencyobservability
◆Current state
Honeybadger's error tracking and Insights query language are mature; the work now is removing the expertise required to use them. Natural language search translates plain English into error filters and BadgerQL, the hosted MCP server accepts browser-approved OAuth instead of hand-pasted credentials, and anomaly detection replaces threshold-tuning with learned baselines. Underneath that, steady platform work continues: EU hosting, S3-compatible archival, Oban-py instrumentation, richer issue exports.
◆Where it's heading
Three consecutive releases each remove a step the user previously had to perform themselves — learn the query syntax, host and credential the MCP server, decide what an alert threshold should be. The pattern points at a product that expects agents and non-experts to be the ones asking the questions, with humans reviewing answers rather than composing queries. Enterprise plumbing is being laid in parallel: EU regions and object-storage archival are procurement answers, not developer features.
◆Prediction
Expect the natural language layer to reach Insights dashboards themselves — generating or editing widgets from a description — and the MCP surface to expand from reading errors toward acting on them, such as resolving or exporting an issue from an agent session.

◆Recent moves

  1. 23d ago

    Natural language searching for Errors and Insights

    ⚡ SPARK

    Plain-English descriptions now compile into error filters and BadgerQL queries, removing the syntax prerequisite that gated Honeybadger's most capable surface. It is the third consecutive release that deletes a step the user used to perform manually, and the clearest statement yet that querying is meant to be conversational.

    View source ↗
  2. 1mo ago

    OAuth support for MCP servers and EU self-hosting

    ⚡ SPARK

    The hosted MCP server gained OAuth and EU regions, so connecting an agent is a browser approval rather than a Docker deployment plus a pasted credential. It sits directly upstream of the natural language work: both reduce what a caller must know before asking Honeybadger a question.

    View source ↗
  3. 1mo ago

    Alerts now support anomaly detection

    Alerts now learn each project's normal error volume and fire when it deviates, instead of relying on a threshold someone guessed. Consistent with the broader push to move judgment out of configuration, though it changes tuning rather than capability.

    View source ↗
  4. 1mo ago

    Oban-py support for Insights and error tracking

    Adds Oban-py support for exception reporting and per-job telemetry into Insights. A real extension of coverage into Python background jobs, and further evidence that Insights is being fed from every runtime Honeybadger touches.

    View source ↗
  5. 1mo ago

    Include more details in GitHub, GitLab, and Jira issue exports

    Issue exports to GitHub, GitLab, and Jira now carry the full markdown or wiki body rather than a summary. Reduces the round trip back into Honeybadger when work is triaged in a tracker.

    View source ↗
  6. 2mo ago

    Archive Insights data in S3-compatible object storage

    Business and Enterprise accounts can replicate Insights data to S3 or R2 for retention beyond Honeybadger's own window. Part of the same enterprise groundwork as EU hosting: an answer to retention policy rather than a new analytical capability.

    View source ↗