Consolidate security audit findings into a deduplicated known-issues.json register.
npx skills add https://github.com/J4X-Security/K.I.T --skill kit-j4x-security
Please help me install this Agent Skill. Skill: kit Source: https://github.com/J4X-Security/K.I.T/tree/main/claude-skill-kit Command: npx skills add https://github.com/J4X-Security/K.I.T --skill kit-j4x-security
KIT solves the problem of repeatedly re-reviewing security audit findings by consolidating prior reports into one canonical known-issues register and then checking whether new findings are already covered.
Ask the AI to build or extend known-issues.json from your audit inputs and then check a new finding against it using the staged workflow.
High-intent search queries and answers about installing and using this skill.
To consolidate duplicate security audit findings, you build a canonical known-issues.json register by ingesting local files, PDFs, URLs, and GitHub repositories. The Skill normalizes extracted content and applies model-driven deduplication to eliminate repeated issues across prior reports.
Yes, you can extract audit findings from PDFs and GitHub repositories. The Skill expands GitHub repo or folder URLs, downloads remote artifacts, and uses pdfplumber and pypdf to extract PDF text before running the staged issue extraction workflow.
The best way to check if a new security finding is already covered is to run a staged extraction step followed by a duplicate-check decision contract. The Skill evaluates the new finding against the full known-issues.json list and returns a verdict on whether it is a duplicate.
Yes, running the security audit deduplication process requires the presence of staged contracts like prepare-build, finalize-build, or prepare-check. These contracts enforce the LLM-authored canonical issue extraction, deduplication, and duplicate-check verdicts against your known-issues list.
Model-driven deduplication handles messy real-world audit inputs by normalizing extracted content from diverse sources like local directories, JSON, and web URLs. The LLM-authored workflow then applies canonical issue grouping to deduplicate extracted issues accurately.
liangdabiaoConvert URLs, PDFs, and local files into an interconnected wiki.
Sampi314Orchestrate 12 audit agents to review Excel financial models and consolidate findings.
ricardoalt1515Extract text and images from PDFs and fill forms using Python libraries.
Rinzler78Review content for weaknesses, omissions, and defects with a Markdown findings list.
