PrizmKit Log Analyzer

Analyze log files to identify anomaly patterns, frequency trends, and error correlations for production issue investigation.

Commands

prizmkit.analyze-logs <log-file-or-directory>

Analyze log files and produce a structured report of findings.

STEPS:

Read and parse log files (auto-detect format):
- JSON structured logs
- Structured text (key=value pairs)
- Syslog format
- Custom formats (infer delimiter and field positions)
Extract entries with normalized fields:
- Timestamp: parse to comparable datetime
- Level: DEBUG, INFO, WARN, ERROR, FATAL
- Source: service name, module, class, or file
- Message: the log message body
- Metadata: request ID, user ID, trace ID (if present)
Analyze patterns:
- Error frequency: count by type and time window (per minute, per hour)
- Correlation: errors that consistently appear together or in sequence
- Anomaly detection: sudden spikes in error rate, new error types not seen before
- Timeline: when did behavior change relative to deployments or config changes
- Request tracing: follow request IDs across log entries to reconstruct flows
Identify top issues:
- Most frequent errors (by count)
- Most recent new errors (not seen in earlier log entries)
- Errors with increasing trend (getting worse over time)
- Errors correlated with specific endpoints, users, or time windows
Generate analysis report:
- Timeline of events: chronological summary of significant changes
- Top 10 error patterns: with frequency, first/last occurrence, and sample messages
- Correlation findings: errors that co-occur or cascade
- Anomaly alerts: unusual patterns that warrant investigation
- Recommended investigation priorities: ranked list of what to look at first

Path References

All internal asset paths MUST use ${SKILL_DIR} placeholder for cross-IDE compatibility.

Output

Structured analysis report printed to console
Summary suitable for sharing with team or pasting into incident reports