feat(security): add Clean-Room Data Hygiene Engineer agent by augustoheiss · Pull Request #1152 · msitarzewski/agency-agents · GitHub
Skip to content

feat(security): add Clean-Room Data Hygiene Engineer agent - #1152

Open
augustoheiss wants to merge 1 commit into
msitarzewski:mainfrom
augustoheiss:feat/add-clean-room-data-hygiene-engineer
Open

augustoheiss wants to merge 1 commit into
msitarzewski:mainfrom
augustoheiss:feat/add-clean-room-data-hygiene-engineer

Conversation

@augustoheiss

Copy link
Copy Markdown
Contributor

New Agent: Clean-Room Data Hygiene Engineer

Division: Security
File: security/security-clean-room-data-hygiene-engineer.md

What This Agent Does

Enforces strict Clean-Room Reverse Engineering and Data/Domain Hygiene at the boundary between user-provided inspiration artifacts (screenshots, scanned PDFs, spreadsheets, legacy operational forms) and generated codebases. When developers use real-world operational documents as visual or structural inspiration for domain-agnostic software, AI coding assistants frequently leak real human names into seed.ts, copy proprietary third-party form codes and verbatim footnotes into built-in templates, ship reference PDFs as downloadable samples, or embed hardcoded lists of real names inside client-side sanitizers (The Sanitizer Paradox). This specialist prevents and remediates all four contamination vectors.

Why It's Needed

Existing Security division agents cover general application security (security-application-security-engineer.md), secrets/API-key hygiene (security-secrets-credential-hygiene-engineer.md), AI-generated code vulnerabilities (security-ai-generated-code-security-auditor.md), and privacy regulations (legal-data-privacy-officer.md). None address AI inspiration bleed-through — where an AI agent extracts real PII, personal names, and source-organization DNA from reference screenshots/PDFs and bakes them into seeds, declarative templates, unit tests, and client storage migrations.

Quality Checklist

  • Follows the agent template structure
  • YAML frontmatter is valid (name, description, color, emoji, vibe)
  • Includes real, working code/deliverable examples (Translation Matrix, Anti-Paradox Client Storage Migration, Deterministic Synthetic Taxonomy Generator, Automated Clean-Room CI Gate)
  • Has specific, measurable success metrics
  • Personality is distinct and consistent throughout
  • Added to README.md under the appropriate division
  • Tested with at least one AI tool (lint-agents.sh 0 errors/0 warnings, check-agent-originality.sh 0.0% similarity, check-divisions.sh passed, installed in Antigravity)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant