Claude Skills for Perl SDET & Architecture
SaferSkills independently audited claude (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Claude Skills for Perl SDET & Architecture
A curated collection of highly optimized, token-efficient system prompts ("skills") designed for Anthropic's Claude LLM. These skills transform Claude into a ruthless, constructive Senior Perl Architect and SDET, capable of automating the entire Perl software development lifecycle--from scaffolding and modernization to pathological edge-case testing and CPAN release auditing.
Designed specifically for the Claude Code CLI (but completely usable in the web UI or other API wrappers), these prompts use strict formatting, agentic instructions, and <thinking> protocols to eliminate AI hallucinations and enforce modern, rigorous Perl standards.
These skills are engineered to bypass common LLM traps:
prove" (which it cannot natively do without terminal tools), the prompts instruct Claude to simulate the execution via static analysis and automatically fix the code if it spots a failure.use strict;, use warnings;, use autodie qw(:all);, strict tab indentation, and ASCII-only code.<thinking> blocks so Claude verifies its logic, identifies dead code, and double-checks POD drift _before_ generating final output.Sub::Private/Sub::Protected encapsulation, resolves TODOs, and synchronizes POD.Params::Validate::Strict and replacing outdated CPAN modules..pm files with full Z-calculus POD schemas.Test::Mockingbird and checking memory cycles.$@, $!, $_).Test::Without::Module).Test::Permissions), hostile encodings (Zalgo/RTL), and rigorously stress-tests for memory exhaustion, Regular Expression Denial of Service (ReDoS), and file descriptor load depletion.TODO markers into the source code for unreachable or dead lines./tmp/ file usage.%ENV variables to test for command injection, traversal, and XSS.TODO markers. Generates a strategic post-release roadmap focusing on new feature ideation and architectural evolution.$VERSION sync, dependency completeness, MANIFEST accuracy, and leftover debugging artifacts.If you are using the Claude Code CLI, you can leverage these markdown files directly by aliasing them or passing them as contextual prompts.
Because Claude Code has file-system access, you can run commands like:
claude -p security-audit.md "Review lib/MyModule.pm"
claude -p cpan-audit.md "Do a final check of this repo"
_Pro-Tip: You can set up custom bash aliases in your ~/.bashrc or ~/.zshrc to make execution instantaneous:_
alias claude-critique='claude -p /path/to/repo/critique.md' alias claude-pentest='claude -p /path/to/repo/pen-test.md'
.md skill file to the project's knowledge base (or paste it into the system instructions).User.pm file."_This toolkit is provided as-is. Feel free to fork, adapt, and integrate into your own CI/CD pipelines and prompt libraries.
Copyright 2026 Nigel Horne.
Usage is subject to the GPL2 licence terms. If you use it, please let me know.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.