lightrun-live-runtime-debugging

Guide deterministic runtime investigations in live environments using Lightrun MCP tools, with preflight gating, recovery/resume rules, evidence-first diagnosis, and explicit blocker/handoff outputs.

1.16x

Quality

36%

Does it follow best practices?

Impact

99%

1.16x

Average score across 3 eval scenarios

Securityby

Advisory

Suggest reviewing before use

Optimize this skill with Tessl

npx tessl skill review --optimize ./skills/lightrun-live-runtime-debugging/SKILL.md

Evaluation results

98%

18%

Checkout Payment Failure Investigation

Investigation framing, hypotheses-first structure, preflight and source selection planning

Criteria

Without context

With context

Hypotheses before tools

100%

Confirms/Weakens per hypothesis

100%

Expected vs unexpected condition

70%

100%

Preflight as first tool step

30%

100%

Source selection reasoning

100%

Tool inspection step

100%

Codepath mapped before actions

100%

80%

Evidence steps tied to hypotheses

100%

Decision per action documented

100%

Investigation template populated

100%

Session Service Investigation: Next Steps

Async activation gate and action lifecycle handoff after failed sync captures

Criteria

Without context

With context

Async mode selected

100%

Activation rule cited

100%

Mode decision summary present

100%

Async action metadata

100%

Long async window used

100%

No premature diagnosis

100%

Bounded in-session polling described

50%

100%

Hits-arrive continuation plan

60%

100%

Handoff includes action IDs

100%

Retry condition specified

100%

14%

Order Service Investigation: Final Diagnosis and Handoff

Cleanup gate review, final diagnosis with evidence traceability, and code-level fix proposal

Criteria

Without context

With context

Cleanup gate present

70%

100%

All actions classified

100%

Correct terminal classifications

40%

100%

Hypothesis results explicit

100%

Bug mechanism: trigger and path

100%

Evidence tied to diagnosis

100%

Fix: file and behavior change

100%

Fix: addresses mechanism

100%

Fix: risk and validation

80%

100%

Confidence and unknowns

70%

100%

Repository: lightrun-platform/lightrun-ai
Commit: c757387

Evaluated: 15 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Checkout Payment Failure Investigation Session Service Investigation: Next Steps Order Service Investigation: Final Diagnosis and Handoff

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.