Bug Recorder

@Bug Recorder log this exception: UserService.authenticate threw NullPointerException when email field was empty, stack trace shows line 247 in auth module

Recorded error in Bug Recorder (ID: err_8k2m9p). The exception has been logged with the provided stack trace details and tagged as a NullPointerException in the UserService.authenticate method. If your account has notification rules configured for authentication errors, those alerts have been triggered. You can view the full error report at bugrecorder.com/errors/err_8k2m9p.

This example shows the core workflow: capturing an error description from conversation and persisting it to Bug Recorder's system. The MCP sends the error immediately — there's no draft or review step. Make sure your API key has write permissions, and be aware that notification rules fire automatically if configured in your Bug Recorder account.

@Bug Recorder record a production incident: payment gateway timeout at 14:32 UTC, affected 23 checkout attempts, Stripe webhook returned 504 errors

Error recorded in Bug Recorder (ID: err_9n3x1q). The incident details — including timestamp, affected user count, and Stripe gateway context — have been logged. Bug Recorder has categorized this as a payment-related timeout. Any team members subscribed to payment error notifications will receive an alert. The error is now visible in your Bug Recorder dashboard under production incidents.

This demonstrates logging a time-sensitive production issue with business impact metrics. The MCP doesn't validate whether the incident actually occurred — it trusts the prompt's description. Use this for real-time incident tracking, but remember that Bug Recorder stores whatever you send, so accuracy in the prompt matters for post-mortem analysis.

@Bug Recorder summarize the last three errors I mentioned in this chat and log a consolidated report titled 'API reliability issues - Jan 15'

I've reviewed the three errors discussed earlier in this conversation: the UserService authentication null pointer, the payment gateway timeout, and the rate-limiting failure on the search endpoint. I've created a consolidated error report in Bug Recorder (ID: err_4p7k2m) titled 'API reliability issues - Jan 15' that synthesizes these incidents, noting the common thread of external service dependencies. The report includes timestamps and affected components for each issue.

This example pairs the AI's conversational memory with Bug Recorder's logging capability. The MCP only sends one error record — the AI generates the summary text from chat history before calling the tool. This is useful for end-of-day incident reviews, but the consolidated report won't link back to individual error IDs unless you explicitly include them in the summary prompt.

When Bug Recorder fits a lean on-call rotation

A 3-person backend team runs on-call rotations and needs to log production errors without leaving Switchy during incident response. Bug Recorder's single-tool MCP lets the on-call engineer record the error, trigger Slack or PagerDuty notifications, and keep working in the same conversation thread. This works when your error-tracking setup is already in Bug Recorder and you're logging 5-15 incidents per week. If you're triaging 50+ errors daily or need to query historical incidents, the MCP is too narrow—you'll spend more time context-switching than you save. For small teams with infrequent but critical production issues, this MCP keeps the incident log in one place without opening a separate dashboard.

Recording escalations when support owns the error log

A 6-person support team escalates customer-reported bugs to engineering through Bug Recorder. The MCP lets support reps log the error directly from a Switchy conversation with the customer, attaching context and triggering the engineering notification in one step. This makes sense when support is the first line of defense and Bug Recorder is the single source of truth for escalations. The trade-off: if your team needs to search past escalations or link errors to tickets in Linear or Jira, the MCP doesn't expose those reads—you'll still open Bug Recorder's UI. For teams where support logs 3-10 escalations per day and engineering picks them up asynchronously, this MCP closes the loop without adding another tool to the support workflow.

When QA needs a write-only error feed

A 2-person QA team runs manual test passes before releases and logs failures in Bug Recorder for the dev team to review. The MCP fits when QA is documenting 5-20 test failures per release cycle and the dev team already monitors Bug Recorder for incoming issues. The single 'Send Error' tool is enough because QA isn't querying or updating errors—they're just recording them with enough context for devs to reproduce. This breaks down if QA needs to check whether an error was already logged or if the dev team wants to assign or close errors from Switchy. For small teams where QA's job is to feed the error log and devs triage elsewhere, this MCP keeps the QA workflow inside Switchy without overbuilding the integration.