Imported from ganeshkumarchalla/Qore (
AGENTS.md). Install upstream withnpx skills add ganeshkumarchalla/Qore. Copyright stays with the author.
QORE Workspace Security & Agent Guidelines
This project implements strict multi-agent security and privacy protocols. All agents, automated workflows, and developer tools operating within c:\Users\ganes\OneDrive\QORE must adhere to these policies.
1. Zero-Exposure Secret Policy
- Never Output Secrets: Never print, log, return, or include contents of
.env, API keys (GEMINI_API_KEY, etc.), bearer tokens, private keys, or passwords in chat responses, error messages, logs, or reports. - Redaction Mandatory: Any tool output or report that touches secrets must display redacted masks (e.g.
AIzaSy...[REDACTED]). - Client/Frontend Isolation: Never place backend API keys or secrets in
frontend/files or prefix them withVITE_. In Vite,VITE_*environment variables are baked into static client bundles sent to browsers. All AI and sensitive requests must route through the backend (backend/qore/).
2. File Safety & Version Control
- Ignore Protection: Ensure
.env,.env.*, certificates (*.pem,*.key), databases (*.db,*.sqlite*), and credential files are strictly listed in.gitignore. - No Accidental Staging: Before committing or sharing files, verify untracked sensitive files are not staged or exposed.
- Temporary & Backup Files: Clean up and ignore editor swap files (
*.swp), backups (*.bak), and system metadata (.DS_Store,Thumbs.db).
3. IDE & Environment Hardening
- Workspace Trust: Maintain active workspace trust (
security.workspace.trust.enabled: true) in.vscode/settings.json. - Search & Index Exclusion: Sensitive files (
.env*,.venv/, credentials) must be excluded from IDE file watchers and search indexes. - Telemetry Disabled: Ensure editor telemetry is disabled to prevent accidental diagnostic data transmission.
4. Multi-Agent Automated Security Auditing
- Run the security audit engine at
scripts/security_checker.pyusing the project's Python virtual environment:& "C:\Users\ganes\OneDrive\QORE\.venv\Scripts\python.exe" scripts/security_checker.py --check - Automated audit reports are generated under
evidence/security_report.mdandevidence/security_report.json.
5. Frontend Quality, UI Resilience & Multi-Agent Guidelines
- All agents and subagents modifying UI, styles, or frontend code must strictly adhere to
docs/FRONTEND_AGENT_GUIDELINES.md. - Run the frontend health audit engine using the project's Python virtual environment:
& "C:\Users\ganes\OneDrive\QORE\.venv\Scripts\python.exe" scripts/frontend_auditor.py - Automated frontend health reports are generated under
evidence/frontend_health_report.mdandevidence/frontend_health_report.json. - Subagents must consult the Frontend Autonomous Agent's standards before modifying components, state structures, or API contracts.
