Index of every page on this site: llms.txt
All skills
launchdarkly avatar

/investigate

@f0ee110 official
Run it yourself

Finds the root cause of a production problem from observability data (logs, traces, errors, sessions, metrics): scope and onset, what changed (deploys and flag changes), which host, version, route or flag variation is over-represented against a baseline, then trace, log and error evidence and blast radius. Use when the user reports a bug, an outage, an error or latency spike, an incident, or asks about patterns across application data.

Use this Skill: https://skilld.dev/gh/launchdarkly/ai-tooling/investigate

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

sessions.md

≈1k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Investigating with sessions

Load this file when the investigation concerns user journeys, frontend behavior, or session replays.

When to reach for sessions

  • The user reports a frontend bug — "checkout broken", "infinite spinner", "empty state"
  • You need to reconstruct what a user saw and did
  • You have a session ID and want to pull events, logs, and errors within that session
  • You're correlating frontend events with backend traces using a shared user identifier or timestamp

Tool guidance (search-sessions)

The search-sessions tool returns sessions with user details and core attributes.

  • startDate is required.
  • query filters by attributes — e.g. identifier='user@example.com' AND has_errors=true.
  • count defaults to 10, max 50.
  • sortField — created_at, length. Defaults to created_at.
  • sortDesc — Boolean. Defaults to true.

Fetching a specific session by secure_id. Use a wide time window (you often don't know when the session happened):

{"projectKey": "my-project", "startDate": "2026-01-01T00:00:00Z", "query": "secure_id=LWK4jxZch4GMmVYJ0rW3R6CNymSy", "count": 1}

Relating events within a session. Session filters like visited-url=*/a* visited-url=*/b* only check that both matched, in any order. To ask about order, absence, or which events happened together, wrap each event in span(...) or log(...) and join the blocks with THEN, AND, or OR, for example span(view_name="/pricing") THEN span(view_name="/checkout"). Load relationship-search.md for the full syntax.

Session IDs in output. When you reference sessions in your response, wrap session IDs in <session-id> tags so the UI can turn them into links:

<session-id>c6wX2TtznM6F0QByEiRBoQBFw1Bv</session-id>

Supporting tools

  • search-timeline-events — pull the event timeline from a session (navigations, clicks, errors, network). Use this to find the exact moment a user hit a problem.
  • search-logs — pass secure_session_id=<id> in the query to get logs scoped to the session. startDate = session created_at, endDate = created_at + 3 hours.

Typical patterns

  1. Bug reproduction — start with the symptom ("infinite spinner on checkout"), find sessions that hit the target URL via query="visited-url=*/checkout*" (or narrow to the failing sequence with query="span(view_name=\"/checkout\") THEN span(has_errors=true)"), use search-timeline-events to walk chronologically and find the last successful action vs. the first failing one.
  2. User-specific investigation — query="identifier=<email>" or by user ID. Pull recent sessions, look for errors or abnormal patterns.
  3. Correlated backend errors — sessions with has_errors=true connect to error groups via the trace_id attribute in session events. Cross-reference.

Interpreting results

  • Walk chronologically. The transition from the last successful action to the first failing one is where the interesting detail sits.
  • Cite exact timestamps. "Around 2:00 into the replay" is less useful than an absolute time.
  • If the session is truncated or missing segments, say so explicitly.
  • Pull errors as a separate query rather than paraphrasing from the session payload — the error group record has the stack trace.

Common mistakes

  • Telling the user that sessions search can't check event order or absence. Use relationship search (relationship-search.md).
  • Putting session keys (visited-url, clickTextContent, event) inside span(...). They parse but match nothing there; use the span keys in relationship-search.md.
  • Forgetting to wrap session IDs in <session-id> tags. The UI can't link them otherwise.
  • Filtering by user identifier without a time window. Returns all sessions ever — expensive and noisy.
  • Using search-timeline-events for overview when you only need a high-level triage read, or skipping it when you actually need chronological detail. Match the tool to the depth you need.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning2d3 checks · Risk SAFE
  • Gen Agent Trust Hub2d

    This skill provides specialized instructions for investigating production outages and performance issues using the LaunchDarkly observability platform. It provides a structured methodology for identifying root causes through log analysis, trace inspection, and error grouping, utilizing the official LaunchDarkly MCP server.

  • Socket2d

    No alerts

  • Snyk2d

    Risk: MEDIUM · 1 issue

Signed by skilld at f0ee110. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 days ago
compatibility
Requires the remotely hosted LaunchDarkly MCP server
metadata
{
  "author": "launchdarkly",
  "version": "0.5.0"
}

README badge

README badge for launchdarkly/ai-tooling/investigate