forensics

Root-cause debugging analyzes system logs to identify patterns in repeated failures. Forensics parses JSONL logs from your execution history, cross-references them with bug records, and surfaces recursion loops, context loss, and environment drift so you can fix the underlying issue.

Pattern tracing identifies recurring bottlenecks and tool usage issues across your session history. By reviewing prompts, contexts, and tool outputs that led to errors, you spot systematic problems—like tools called with inconsistent parameters or environments drifting between runs.

Cognitive audits review the prompts and execution contexts that preceded failures. This traces error sources back to specific decisions or state changes, helping you understand whether problems stem from prompt formulation, missing context, or environmental shifts.

Analyzing your last failure isolates the most immediate problem. Forensics examines the final failure's logs and context to clarify whether the issue is reproducible, environmental, or tied to a specific tool invocation, guiding your remediation approach.

Forensics processes JSONL logs, execution records, and bug metadata stored in your History and Execution directories. It detects recursion loops, context loss, and environment drift by cross-referencing raw outputs with failure records to pinpoint systemic causes.