skill-creator

Session pattern detection scans local or global pi agent session history to identify repetitive workflows, bash command sequences, and prompt patterns occurring three or more times. It analyzes your history filtered by current working directory or across all projects to surface consistent multi-step workflows.

Creating custom skills from repeated bash commands involves a guided process that drafts required YAML frontmatter, writes operational instructions, validates skill structure, and places the skill in the correct local or global directory for immediate activation.

Yes, jq is a required dependency to analyze session files and generate custom skill configurations. This lightweight command-line JSON processor is used by the skill's internal scripts to parse session history and detect repetitive patterns.

You can detect command sequences in both your current working directory and across all projects. The session analysis applies to local session history filtered by the current directory and global session history across all projects to find repeated workflows.

Guided skill creation validates YAML frontmatter and directory placement by checking the structural formatting and confirming the correct local or global folder location. This validation ensures the custom skill is properly configured before it becomes active.

A limitation of detecting prompt patterns is that the analysis only identifies repeated sequences occurring three or more times. Workflows with minor variations or low frequency may not be recognized, requiring manual review to capture less consistent session patterns.