debug

Investigate stuck runs and execution failures by tracing Symphony and Codex logs with issue/session identifiers; use when runs stall, retry repeatedly, or fail unexpectedly.

1.00x

Quality

88%

Does it follow best practices?

Impact

83%

1.00x

Average score across 3 eval scenarios

Securityby

Passed

No known issues

Evaluation results

69%

Stuck Run Investigation

Log investigation script and report for stuck run

Criteria

Without context

With context

Uses rg not grep

rg -n flag

Searches log/symphony.log*

issue_identifier first

50%

75%

session_id extraction

60%

50%

Traces session by session_id

62%

87%

Lifecycle reading

100%

Failure classification

100%

Exact failing stage

100%

Evidence with correlation fields

100%

90%

Multiple sessions noted

100%

Stall/backoff signals cited

100%

Run Failure Incident Report

Incident report for repeated app-server startup failure

Criteria

Without context

With context

App-server startup classification

100%

Pre-stream failure stage

100%

All three session IDs identified

100%

issue_id paired with findings

100%

issue_identifier in evidence

100%

session_id in evidence

100%

Timestamps in evidence

100%

Scope: isolated to MT-714

100%

Scheduling retry signals cited

100%

Root cause recorded

100%

No mixing concurrent sessions

100%

82%

Worker Failure Pattern Analysis

Multi-ticket scope analysis for worker crash pattern

Criteria

Without context

With context

Uses rg not grep

rg -n flag used

Failure signal search

100%

Worker crash classification

100%

Agent task exited signal cited

100%

Systemic conclusion

100%

Successful tickets noted

100%

Both failing tickets covered

100%

issue_identifier in evidence

100%

session_id in evidence

100%

Timestamps in evidence

100%

Searches log/symphony.log*

42%

Repository: openai/symphony
Commit: 4cbe3a9

Evaluated: 13 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Stuck Run Investigation Run Failure Incident Report Worker Failure Pattern Analysis

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.