The Agent Skills Directory

[SAFE]: The skill implements a methodology for 'Session Archaeology', which involves parsing and analyzing log files from previous AI sessions. The provided Python implementation uses standard libraries for data processing and does not perform any dangerous network operations, file modifications, or unauthorized code execution.
[DATA_PROCESSING]: The skill defines structures for parsing JSONL session transcripts. While it reads external files, it does so to extract statistics and classify events (tool calls, errors, etc.) without executing the content of those files.
[INDIRECT_PROMPT_INJECTION]: The skill inherently processes untrusted data (session transcripts which contain user inputs). While this represents a theoretical attack surface for indirect prompt injection if the analysis results were automatically fed back into a model's system prompt without sanitization, the skill itself only provides diagnostic tools and taxonomy builders, which is a safe practice.

session-archaeology