Personal collection of Claude agents and commands.

agents/                # Specialized agents (architect, debugger, developer, etc.)
commands/              # Commands like plan-execution
prompt-engineering.md  # Prompt optimization techniques

Launches a security audit estimation workflow that orchestrates parallel sub-agents for LOC analysis, documentation review, and complexity scoring before synthesizing a precise effort estimate.

A lighter-weight audit estimator that still runs parallel LOC sweeps per folder and a single categorization pass, ideal when you want fast refined/raw line counts without the extended documentation and complexity analysis.

Puts you in a project-manager role that delegates tightly scoped tasks to specialized agents, tracks progress with TodoWrite, and enforces validation after every incremental change.

Bootstraps a full audit engagement by coupling the parallel LOC foundation with five deep-dive analysis agents (component summaries, entry points, file categorization, critical functions, security operations) to output an end-to-end kickoff report with priorities, effort allocation, and methodology guidance.

prompt-engineering.md is extracted from Southbridge Research's exceptional analysis of Claude Code prompts. This document has been invaluable for understanding and applying advanced prompt engineering techniques.

When optimizing prompts with these techniques:

Apply ALL prompt engineering patterns from `prompt-engineering.md` to this modification. For EACH change, specify EXACTLY which technique(s) you used:
- Pattern name (e.g., "Progressive Disclosure", "Emphasis Hierarchy")
- Why this pattern applies here
- Expected behavioral impact

CRITICAL: Changes without pattern attribution = task incomplete.