skill-session-analyzer

To verify Claude Code session logs against a SKILL.md specification, you provide a session ID and target SKILL.md path. The analyzer extracts expected behaviors, parses debug log traces for subagent and hook events, and generates an expected-vs-actual PASS/FAIL report.

Yes, you can audit hook execution and subagent workflows from past debug logs. The analysis extracts subagent invocations, hook matches, permission decisions, and tool execution evidence from session traces to identify workflow sequencing deviations.

Post-hoc session validation for Claude Code is the process of comparing expected workflow elements from a target SKILL.md against actual recorded behavior in session logs. It catches mismatched workflows, missing hook calls, and unexpected artifact behavior after execution.

To check if expected files were created during a Claude Code session, the analyzer verifies filesystem artifacts by searching tracked file history and atomic write signals. It validates whether expected files were created, written, or deleted according to the SKILL.md specification.

No, you do not need to manually locate session files under the ~/.claude directory to audit them. You simply provide the session ID and the path to the target SKILL.md, and the analyzer automatically locates and parses the required session files.

Missing hook calls in Claude Code subagent workflows appear in debug logs when actual execution deviates from the expected SKILL.md specification. The analyzer identifies these mismatches by parsing log traces for hook matches and permission decisions to produce a structured deviation report.