Add Sentry Error Intelligence dashboard with Slack search

[liamdebeasi]

Summary

Adds a new "Sentry Error Intelligence" dashboard that surfaces top errors, escalating issues, and related error groups. Enhances the Slack integration with thread enrichment, fuzzy search, and user token support, and extends the Sentry data source to require an org slug.

Problem

There was no dedicated app for analyzing Sentry errors beyond what the official Sentry UI provides. Additionally, the Sentry data source lacked an org slug configuration field, and the Slack search integration only supported a bot token (which cannot use the search.messages API) and did not surface thread context or clickable message links.

Key Changes

• New sentry-errors dashboard registered in the adhoc dashboard registry (ErrorGroupsPanel.tsx and related components) with top-10 error analysis, escalation detection, related error grouping, and project filtering
• Sentry SENTRY_ORG_SLUG data source field added to the walkthrough steps so users can configure their org slug via the UI
• getCodeMappings added to the Sentry server lib and exposed as a new code-mappings mode in the Sentry action
• github-blame action added to fetch Git blame for a file path via GitHub's GraphQL API
• Slack SLACK_USER_TOKEN data source field added as an optional walkthrough step (with an "Add" button replacing the static "Optional" label in the connected view)
• Slack searchMessages overhauled: prefers a user token (xoxp-) for search.messages; falls back to bot-token channel history scanning with fuzzy matching (URL normalization, numeric suffix matching, prefix matching)
• Thread enrichment (enrichWithThreads) fetches full reply threads for matched Slack messages so conversation context is surfaced alongside search results
• Slack search limit passed through from the action call; results capped and sorted by recency
• User resolution fallback removed fake placeholder entries so resolveUsername can fall back to the username field populated by the search API

---

  

---

To clone this PR locally use the Github CLI with command gh pr checkout 1035

You can tag me at @BuilderIO for anything you want me to fix or change

[builder-io-integration]

Builder reviewed your changes and found 5 potential issues 🔴

Review Details

Code Review Summary — PR #1035: Sentry Error Intelligence Dashboard

This PR adds a new "Sentry Error Intelligence" dashboard to the analytics template with rich error analysis, escalation detection, and enhanced Slack/GitHub integration. The feature includes:

• New sentry-errors dashboard with top-10 error analysis, escalation detection, and related error grouping
• Sentry org slug support via new data source field for proper multi-org configuration
• Enhanced Slack search with optional user-token support and intelligent fallback (fuzzy search with bot token)
• GitHub blame integration exposing commit history and contributor context for errors
• Thread enrichment fetching full Slack conversation context alongside search results

Architecture Assessment: The implementation is well-structured with proper separation of concerns (UI components, actions, server-side integrations), appropriate credential handling, and good error fallback patterns. Type safety is solid throughout.

Risk Classification: Standard risk (analytics feature addition, no breaking API changes, no database schema modifications).

Key Findings

Critical Issues: 🔴

• Missing function import breaks Slack user-token search (ReferenceError at runtime)

Should-Fix Issues: 🟡

• GitHub blame feature shipped but not integrated into the dashboard UI
• Slack secondary-workspace searches use wrong user token credential
• Sensitive Slack search content and queries leaked to server logs
• Slack URL matching has trailing-slash normalization issue
• Unused Sentry stats query adds unnecessary API overhead

Positives: ✅

• Proper credential scoping and access control
• Well-designed async error handling and fallbacks
• Correct React patterns (memoization, hook dependencies)
• Thoughtful API integration (Slack, GitHub, Sentry)

🧪 Browser testing: Will run after this review (PR touches UI code)

[builder-io-integration]

🔴 ReferenceError: resolveCredential not imported in searchMessages

The searchMessages function calls resolveCredential("SLACK_USER_TOKEN", ctx) on line 335, but this function is no longer imported. The import was changed to only resolveAnalyticsProviderCredential. This will cause a ReferenceError at runtime when Slack user-token search is attempted, crashing the entire search flow. Either reimport resolveCredential from "./credentials" or replace the call with resolveAnalyticsProviderCredential().

_{Or react with 🚀 to auto-fix.}

[builder-io-integration]

🟡 GitHub blame feature not mounted in the dashboard

The PR adds GitHubBlamePanel component and github-blame action, but IssueDetail only renders SlackMentionsPanel. The new GitHub blame feature is unreachable from the dashboard, making it a non-functional addition. Import and render GitHubBlamePanel alongside the Slack panel to complete the feature.

_{Or react with 🚀 to auto-fix.}

[builder-io-integration]

🟡 Secondary workspace uses same user token as primary

The searchMessages function prefers SLACK_USER_TOKEN for both primary and secondary workspaces. Since only one user-token credential was added (no SLACK_USER_TOKEN_2), secondary workspace searches will query the wrong Slack workspace when a user token is configured. Add per-workspace user-token support (SLACK_USER_TOKEN_2) or disable the user-token path for secondary workspace queries.

_{Or react with 🚀 to auto-fix.}

[builder-io-integration]

🟡 Sensitive Slack content leaked to server logs

The fallback Slack search path logs raw search queries, channel names, and the first 80 characters of matched message text (lines 366, 393, 436). This exposes sensitive workspace content to server logs and any connected observability provider. Remove these logs or gate them behind a production-disabled debug flag. Never log raw message snippets.

_{Or react with 🚀 to auto-fix.}

[builder-io-integration]

🟡 Slack URL search missing trailing-slash normalization

The Slack search query includes the Sentry issue permalink, but the fuzzy matcher in slack.ts explicitly rejects URLs unless they match exactly. If the URL has trailing-slash variations (e.g., ...issues/123 vs ...issues/123/), the search will fail silently. Normalize the permalink URL to a canonical form before adding it to the search query (strip trailing slashes consistently).

_{Or react with 🚀 to auto-fix.}

[builder-io-integration]

Browser testing: 13/15 passed, 2 failed

Test Results: 13/15 passed ⚠️

✅ TC-01: Sentry Error Intelligence dashboard loads at correct URL (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-02: Period filter tabs (24h/7d/14d/30d) are visible and can be switched (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 3 screenshots captured

✅ TC-03: Four stat cards are visible with correct labels (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-04: Correct error state is displayed when no Sentry credentials configured (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 1 screenshot captured

✅ TC-05: Dashboard is listed in sidebar and sidebar navigation works (succeeded)

URLs tested: http://127.0.0.1:8080/analytics, http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-06: Issues list renders correctly with mock data (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-07: Clicking an issue expands the detail view with metadata (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-08: Sparkline charts render in issue rows when stats data is present (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-09: GitHub blame panel is NOT rendered in the issue detail view (known bug verification) (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-10: Issue classification badges are shown correctly for different error types (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 1 screenshot captured

✅ TC-11: Error Groups tab groups issues by error type with expand/collapse (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 3 screenshots captured

❌ TC-12: Slack mentions panel renders in issue detail and initiates search (failed)

Steps: Navigated to Sentry dashboard with mock issues, expanded the first issue detail, found the pre-filled Slack query button, clicked it, waited, inspected DOM for visible changes, checked console for errors.

Failure: assertion_failed — After clicking the prefilled Slack search query button, the Slack mentions panel showed no visible response — no loading skeleton, no error message, no empty state. Expected: a user-friendly result (e.g., 'Slack is not connected' or 'No messages found'). The mutation may be silently failing or the response (likely 'missing_api_key' error code) is being shown in a non-visible part of the UI. No ReferenceError/crash was observed.

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 3 screenshots captured

❌ TC-13: Data sources page shows Sentry "Not configured" and opens config dialog (failed)

Steps: Navigated to /analytics/data-sources, found Sentry under ENGINEERING section (Not configured), clicked it to open config flow, advanced to auth token step, searched DOM text for SENTRY_AUTH_TOKEN string.

Failure: assertion_failed — The Sentry configuration dialog under Data Sources shows "Enter your Auth Token" with placeholder "sntrys_..." but never mentions the environment variable name SENTRY_AUTH_TOKEN. The dashboard error state explicitly says "Add SENTRY_AUTH_TOKEN in Settings → Data sources", creating a terminology mismatch. The config panel does appear and accept a token correctly — just lacks the env var label.

URLs tested: http://127.0.0.1:8080/analytics/data-sources

Evidence: 3 screenshots captured

✅ TC-14: Escalating tab shows correct empty state when no escalating issues (succeeded)

URLs tested: http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 2 screenshots captured

✅ TC-15: Other adhoc dashboards still work after registry change (regression) (succeeded)

URLs tested: http://127.0.0.1:8080/analytics, http://127.0.0.1:8080/analytics/adhoc/test-dashboard-b-1778877310827, http://127.0.0.1:8080/analytics/adhoc/explorer, http://127.0.0.1:8080/analytics/adhoc/sentry-errors

Evidence: 4 screenshots captured

Details

PR #1035 Sentry Error Intelligence Dashboard: 13/15 tests passed. Core navigation, period tabs, stat cards, error states, issue display, sparklines, classification badges, escalation detection, Error Groups tab, and regression coverage all passed. Two failures: (1) Slack mentions search showed no visible feedback after activation — the search button click did not surface any result/error state in the UI; (2) Sentry config panel does not mention SENTRY_AUTH_TOKEN by name, creating a UX inconsistency with the dashboard error message. Additionally, TC-09 visually confirmed the known PR bug: GitHubBlamePanel.tsx exists but is never imported or mounted in the dashboard.